The Bomb: Presidents, Generals, and the Secret History of Nuclear War

by Fred Kaplan

In public, over the years, officers and officials have described America’s nuclear policy as second-strike deterrence: if an enemy strikes us with nuclear weapons, we will retaliate in kind; this retaliatory power is what deters the enemy from attacking us. In reality, though, American policy has always been to strike first preemptively, or in response to a conventional invasion of allied territory, or to a biological or large-scale cyber attack: in any case, not just as an answer to a nuclear attack. All of these options envision firing nuclear weapons at military targets for military ends; they envision the bomb as a weapon of war, writ large. This vision has been enshrined in the American military’s doctrines, drills, and exercises from the onset of the nuclear era through all its phases. Most presidents have been skeptical of this vision—morally, strategically, practically—but none of them have rejected it. Some have threatened to launch nuclear weapons first as a way of settling a crisis. The few who considered adopting a “no first use” policy, in the end, decided against it.

In 1947, in recognition of its impact on the recent victory, the Army Air Forces were declared a separate service, coequal with the Army and the Navy. This was a significant move. As long as they were a unit within the Army, the air forces would support the Army’s policies and missions, mainly by providing air support to troops on the battlefield. Even LeMay’s bombing raids were initially intended to degrade Japan’s military machine and thus pave the way for the Army’s impending invasion of the mainland. But as a separate service, the U.S. Air Force, as it was now called, could set its own missions and strategies.

At first, the Army and the Navy tried to halt history too. In 1949, when the Defense Department cut the budget for ships in order to buy more bombers, the Navy’s top echelon of admirals staged an unprecedented revolt. Several of them condemned the A-bomb on moral grounds. At a congressional hearing, Rear Admiral Ralph Ofstie—who had served on the U.S. Strategic Bombing Survey, a postwar panel of officers and economists that downplayed the role of air power in the Allied victory—condemned Air Force–style city-bombing as “random mass slaughter,” “ruthless and barbaric,” and “contrary to our fundamental ideals.”

By 1952, a mere three years after the admirals’ revolt, all of the Navy’s new combat planes were designed to carry nuclear weapons.

The American military wove the new weapons into its war plans with no hesitation. In March 1954, the Joint Chiefs of Staff—the top officers of all the branches of the armed forces—declared in a Top Secret document, “In a general war, regardless of the manner of initiation, atomic weapons will be used at the outset.” The term “general war” was defined as an armed conflict that pitted American and Soviet forces directly against one another. In other words, a war between the world’s two new superpowers would be, from the first salvos, a nuclear war. This would be the case “regardless of the manner of initiation.” Any armed Soviet incursion into territory deemed vital to U.S. interests—even a tentative crossing of the East-West German border—would spark an instant, all-out nuclear response.

When LeMay had been a colonel, placed in charge of an air division in Europe in the early years of World War II, he heard reports that many pilots were aborting their missions in the face of heavy fire from German fighter planes or antiair batteries. Upon taking command, he told his airmen that he would ride in the lead plane on every bombing run, and if any plane behind him veered away from battle, its entire crew would be court-martialed. The abort rate plummeted overnight.

Vandenberg described his plan as “killing a nation,” but he parsed the campaign into three categories of targets, which he labeled “Delta-Bravo-Romeo.” Delta stood for “the disruption of the vital elements of the Soviet war-making capacity.” Bravo called for “the blunting of Soviet capabilities to deliver an atomic offensive against the United States and its allies.” Romeo stood for “the retarding of Soviet advances into Europe.” In this scheme, destroying all three sets of targets would destroy the Soviet Union as a war-making power. LeMay thought this was abstract nonsense. The quickest way to destroy the Soviet war machine was to destroy the Soviet Union—especially its large cities, where the political leaders and military commanders and factory workers lived.

The work took on a self-serving circular logic: more weapons drove the need to find more targets; more targets propelled a need to buy more weapons. From 1949, the year of SAC’s first war plan, to the spring of 1955, the list of targets grew fourteen-fold—and continued to grow each year through the end of the decade.

Since the collapse of their revolt at the start of the decade, the Navy’s admirals had never stopped resenting the Air Force for usurping the defense budget—nor had they stopped looking for a path to regain their own dominance. In 1956, Admiral Arleigh Burke, the newly named chief of naval operations, thought he found it. Burke, the commander of a carrier task force in the Second World War, had been a keen supporter of a controversial program in the late 1940s to build a nuclear-powered submarine. Now, as the Navy’s top admiral, he jump-started funding for a similar, more modern sub capable of carrying sixteen long-range ballistic missiles, each tipped with a half-megaton nuclear warhead. The missile would be called the Polaris.

NAVWAG’s main thesis was that, once the Soviets had their own nuclear arsenal, the notion of winning a nuclear war was absurd and any plan to launch an American first strike was suicidal. Therefore, the only sensible purpose of nuclear weapons was to deter an enemy attack. The only way to deter an enemy attack was to develop a nuclear arsenal that could answer an enemy attack with a devastating counterblow to the enemy’s cities. The best way to do that was to maintain an invulnerable second-strike force. And the most invulnerable force imaginable was a fleet of missile-carrying submarines moving undetected beneath the ocean’s surface.

The NAVWAG report, and Burke’s whole concept for the Polaris, rested on the premise that a “second-strike” deterrent was enough to keep the nation secure. The Air Force had contributed to this notion, White wrote in a memo to all of its commanders, “through the over-use of the term ‘retaliation.’ ” It was time to emphasize, up front, that in a confrontation with today’s enemy, the nation needed to “take the initiative.” The ability to respond to a Soviet nuclear attack was necessary but insufficient. We also needed the ability to preempt that attack before it happened—to destroy the Soviet missiles and bombers before they were launched, before they inflicted catastrophic damage on American soil.

LeMay had a case, but by early 1960 General White concluded that a unified command was a pipe dream: the admirals and the Army generals were implacably opposed, and for good reason. So White devised what he described in private memos as a “fallback position.” He proposed the creation of a new organization called the Joint Strategic Target Planning Staff. It would consist of officers, from all the services, who would combine their individual nuclear war plans—their lists of what weapons would be fired at what targets—into a Single Integrated Operational Plan. Sitting on top of all these creations—which soon became known by their acronyms, the JSTPS and the SIOP—would be the Director of Strategic Targeting. The catch was that the JSTPS would be located inside SAC headquarters, and the Director of Strategic Targeting would be the same four-star Air Force general who served as SAC commander. In short, the SIOP would be a multiservice war plan, but SAC would plan the war.

At the first SIOP Planning Conference, chaired by General Power on August 24, the main topic on the agenda was the creation of a National Strategic Target List, abbreviated as the NSTL. The Navy officers at the meeting tried to cap how many targets the United States needed to hit with nuclear weapons in the event of war. But Power noted that the NSTL wasn’t about military requirements. The point was simply to list all the objects, all the possible targets in the USSR, the Warsaw Pact nations, and China that had some strategic value—air bases, naval ports, missile sites, radar stations, command-control facilities, military factories, civilian factories. It didn’t necessarily mean that the SIOP would call for all those targets to be attacked. The Navy backed down. As it turned out, the list identified about 4,000 points on the map as “targets.” All told, the Air Force and Navy combined had 3,423 nuclear weapons with the range to hit those targets. It was no coincidence that those two numbers weren’t far apart—or that the number of targets slightly exceeded the number of weapons. If anyone argued that SAC had too many nuclear weapons, General Power could respond that, to the contrary, he didn’t have enough weapons to hit all of the enemy’s targets. The NSTL became a device to justify the arsenal’s expansion.

Calculating the ratios across the board, they saw that the SIOP assigned an average of 2.2 nuclear weapons—each unleashing an explosive force of several megatons—to each and every target across the Soviet Union, China, and Eastern Europe. And, in keeping with LeMay’s philosophy of modern war, these weapons would be launched in waves as massive and rapid as possible. If a war started with short notice, the SIOP called for firing 1,459 nuclear weapons—all of the weapons that were on day-to-day alert, ranging in power from 10 kilotons to 23 megatons, totaling 2,164 megatons in all—against 654 targets. If the United States had time to prepare a first strike, which most SAC officers assumed would be the case (they assumed no president would sit back and let the Soviets or Chinese get in the opening salvo), the plan called for launching 3,423 nuclear weapons—a total of 7,847 megatons—against 1,043 targets.

Before the trip, Rathjens asked a CIA liaison for the name of the Soviet city that most resembled Hiroshima in size and industrial concentration. When he got to SAC headquarters, he asked a staff officer how many weapons the SIOP laid down on that city. The answer: one 4.5-megaton bomb, followed by three 1.1-megaton bombs, in case the first one was a dud. The A-bomb that destroyed one third of Hiroshima, at the end of World War II, had released an explosive force of 12.5 kilotons. In other words, SIOP-62 would hit that Soviet town with more than six hundred times the blast power of the Hiroshima bomb.

General Lemnitzer replied with a brief memo, speaking on behalf of all the Chiefs, dismissing the whole notion of “controlled responses” and “negotiating pauses” in a nuclear war. First, he argued, it was impractical. U.S. nuclear forces were too vulnerable to risk holding back a large number of them once a war had begun; a Soviet strike would disable this so-called reserve force, so, as a practical matter, we had to use it or lose it. Second, there was a conceptual problem. The alleged advantages of “limiting” our attack were plausible only if the Soviets cooperated—that is, if they limited their attack too—and that premise, he wrote, “does not now appear realistic.”I Finally, Lemnitzer argued, the act of even advocating this “controlled response” doctrine “could gravely weaken the current deterrent posture.” Though he didn’t bother to elaborate, his point was clear: to the Chiefs, and to SAC, the appeal of nuclear weapons—the crux of their power to keep the enemy at bay—was precisely their massive destructiveness. Talking about restraining this power, much less touting restraint as a new “doctrine,” might tempt our enemies to flex their muscles without fear of a devastating counterpunch.

Similarly U.S. nuclear responses would be controlled, enemy cities would be excluded from an attack, and a secure reserve of weapons would be retained “to the degree practicable.” What “military necessity” permitted, or what was deemed “practicable,” were matters to be decided by SAC and the Joint Strategic Target Planning Staff, not by McNamara and his whiz kids or even by the Joint Chiefs of Staff. And they decided that these notions were not remotely practicable or consistent with military necessities. Finally, the “damage-expectancy” numbers—the requirement that certain targets be damaged with 90 or 95 or even 97 percent probability, which provoked such agitation in the debates over the first nuclear war plan—remained the same in this revised plan. As in the first war plan, the Navy’s delegates to SAC led the protest against these requirements. The plan, they said, called for too much damage against too many targets, “far in excess” of the Joint Chiefs’ requirement. For all practical purposes, they concluded, SIOP-63 was not much different from SIOP-62. It was still, by and large, a Single Integrated Operational Plan—an only slightly less rigid blueprint for an all-out nuclear first strike that would kill hundreds of millions of people.

Ten days into his presidency, Bundy wrote a memo, laying out the decisions that Kennedy would soon have to make on issues spanning “the whole spectrum from thermonuclear weapons systems to guerrilla action and political infiltration.” On the question of nuclear war, he would need to decide whether to emphasize “strike first” “counterforce” strategies or a “second strike” “deterrent” posture. “The matter is of literally life-and-death importance,” Bundy wrote, “and it also has plenty of political dynamite in it.” The same day, David Bell, the budget director, wrote a memo noting that the three military services disagreed on these strategies. The Air Force advocated “counterforce deterrent,” requiring more heavy bombers and ICBMs; the Navy touted “finite deterrent,” accomplished by a limited number of submarines; and the Army coined the phrase “credible deterrent,” defined as more conventional forces—tanks, troops, artillery, and so forth—to minimize the chance that a small conflict might escalate to nuclear war in the first place.

The intelligence since the Discoverer flights revealed that the Soviets had only a few crucial targets to hit. They included forty-six bases for the Soviet air force’s nuclear-armed bombers, twenty-six staging bases (where Soviet planes would stop for refueling on their way to the United States), and sixteen Soviet ICBM sites (with two “aim points” for each site, to maximize the chance of destroying them), for a total of eighty-eight targets in all.

Kennedy had heard from various aides and officials—McNamara, Taylor, and Bundy in particular—that the SIOP was atrocious and inflexible, and it lived up to its reputation. Hardly a war plan in the traditional sense, it would unleash the same global catastrophe as predicted by the Net Evaluation Subcommittee. “General,” Kennedy asked at one point during the briefing, “why are we hitting all those targets in China? As I understand this scenario, this war didn’t start there.” “They’re in the plan, Mr. President,” Lemnitzer matter-of-factly replied.

After October 1961, Khrushchev saw that he needed real military leverage if he was to make another play for Berlin. American military forces surrounded the Soviet Union, and now he knew that Kennedy knew that the Kremlin had very few—really, next to no—weapons with the range to hit U.S. territory in response to an American first strike. Despite the rhetoric about sausages, the Soviet ICBM program was in doldrums. But Khrushchev did have a fair number of medium-range missiles, and the idea struck him that placing some of those missiles in Cuba—the revolutionary island nation that had recently declared itself a Soviet ally and that sat just ninety miles off American shores—would have the same effect as a crash buildup of ICBMs.

Kennedy cut them off. The real question, he said, is what the Soviets would do if we gave them a chance to withdraw the missiles from Cuba. For instance, he proposed, we could tell Khrushchev, “If you begin to pull them out, we’ll take ours out of Turkey.” This was the first time anyone around the table had mentioned the medium-range nuclear missiles—known as Jupiters—that the United States had recently installed in Turkey, poised on the southern border of the Soviet Union. Kennedy fixed on the idea. Wondering aloud why Khrushchev had taken this wild gamble of putting missiles in Cuba, he figured that they must be part of some diplomatic bargaining scheme and that, to get rid of them, he might have to let Khrushchev save face. “The only offer we would make, it seems to me, that would have any sense of giving him some out,” he told the group, “would be our Turkey missiles.”

Taylor then noted that the crisis offered an opportunity to grab Cuba, a notion that he assumed would appeal to Kennedy a year and a half after the failed Bay of Pigs invasion. “We can never talk about invading again after they get these missiles,” Taylor warned, “because they’ve got these pointed at our head.”

On one point, Kennedy would have agreed with LeMay. He too had come to realize that there was no such thing as a “surgical strike.” A big attack would be needed. But this was precisely why Kennedy disagreed with LeMay on policy. Because of the infeasibility of a limited attack, LeMay wanted an all-out war, while Kennedy wanted to avert one.

After indulging another round of objections from his advisers, Kennedy made his point more firmly. “Now let’s not kid ourselves,” he said. “Most people think that if you’re allowed an even trade, you ought to take advantage of it.” Bundy protested with unusual vigor. “I think we should tell you the universal assessment of everyone in the government who’s connected with alliance problems,” he said, his voice quivering. “If we appear to be trading the defense of Turkey for the threat in Cuba, we will face a radical decline.” Kennedy asked, “How much negotiation have we had with the Turks?” Rusk replied, “We haven’t talked with the Turks.” The president was floored. “I’ve talked about it now”—a possible Cuba-for-Turkey missile trade—“for about a week.” Yet no one had asked the Turks what they think? A few minutes later, Kennedy asked how many missiles we had in Turkey anyway. Nobody knew. An aide in the back of the room called out, “We have fifteen Jupiters.” “Fifteen Jupiters in Turkey?” Kennedy said, once again—judging from his tone—astonished, in this case by the willingness of his advisers to risk so much for the sake of so little.

Not long after, away from the Cabinet Room, President Kennedy sent his brother to the Soviet embassy with a message for Ambassador Anatoly Dobrynin: he would accept the trade, but only if it is never publicly revealed. (Kennedy apparently agreed with Bundy that an open trade might irritate the NATO allies.) The official announcement, Bobby told Dobrynin on his brother’s behalf, must be worded along the lines of Khrushchev’s offer Friday night: a withdrawal of the Soviet missiles in exchange for an American promise never to invade Cuba. Kennedy seemed to think this pledge would be enough to let Khrushchev claim victory to his critics in the Kremlin. (It wasn’t; Khrushchev was ousted two years later, in part, as his successors put it, for his “hare-brained scheme” in Cuba.)

The other palace historians picked up on the same myth—as did the anti-Kennedy revisionist historians, who would cite Kennedy’s alleged rejection of the trade as an example of his dangerous machismo. A quarter-century after the crisis, McGeorge Bundy admitted in his memoir that hushing up the missile trade produced pernicious consequences. “We misled our colleagues, our countrymen, our successors, and our allies” into believing “that it had been enough to stand firm on that Saturday.” Among the misled advisers was Kennedy’s vice president, Lyndon Johnson, who was never told the true story about the trade and who followed the false lesson of the Cuban missile crisis—stand firm, never negotiate—in his decision to escalate the war in Vietnam. Meanwhile, in the autumn of 1962, Kennedy came away from his tangle in two near-catastrophic crises with the conviction that he had to find a way out of the nuclear trap—and, if possible, out of the whole Cold War.

With that, Kennedy summarized the basic dilemma of nuclear strategy: a minimal arsenal of city-busting weapons might be adequate to deter a nuclear attack—this was the Navy’s strategic rationale for the Polaris—but, “if the deterrent fails,” if the threat had to be carried out, no president would want to attack Soviet cities, kill millions of their people. On the other hand, the alternative—destroying the Soviet missile sites, the essence of the counterforce strategy—would require matching the Soviet buildup, which might spur them to build more weapons still, which would require another buildup to match that buildup, and onward and upward the spiral would go.

“You must tell no one outside this suite what Secretary McNamara has told you,” Yarmolinsky said. If the Chiefs or the allies or Congress knew he felt this way, they would panic, and pressure would build for his dismissal. Neither of them had to spell out the reason for discretion. It was the unquestioned premise of Cold War policy that deterrence required persuading the Soviets that the American president would use nuclear weapons first, in response to aggression against U.S. allies. Once that premise was accepted, the logical corollary became hard to resist: to persuade the Kremlin that he would use nuclear weapons first, the president had to develop the ability and display the will to use them first.

On the other hand, Taylor went on, the Air Force did strongly support a “first-strike capability,” which could limit damage to the United States, to levels that were “acceptable in light of the circumstances and alternatives.” He did not define “acceptable,” but the implication was clear: the resulting death toll would be very high—in the hundreds of thousands or millions—and, to the Chiefs, that would be acceptable, given the “circumstances” (nuclear war) and the “alternatives” (surrender).

Kennedy emphasized that he was no pacifist or idealist. “I am not referring to the absolute, infinite concept of peace and good will of which some fantasies and fanatics dream,” he said. “Let us focus instead on a more practical, more attainable peace—based not on a sudden revolution in human nature but on a gradual evolution in human institutions, on a series of concrete actions and effective agreements which are in the interest of all concerned.” With such a peace, he allowed, “there will still be quarrels and conflicting interests.” And he condemned Communism as “profoundly repugnant,” a “negation of personal freedom and dignity.” But, he said, “we can still hail the Russian people for their many achievements” and recognize our “mutual abhorrence of war,” recalling that “no nation in the history of battle ever suffered more than the Soviet Union suffered in the course of the Second World War,” and noting, “Almost unique among the major world powers, we have never been at war with each other.” Yet both nations were caught up in “a vicious and dangerous cycle in which suspicion on one side breeds suspicion on the other, and new weapons beget counter-weapons.” Both sides, then, “have a mutually deep interest in a just and genuine peace and in halting the arms race.” So “let us not be blind to our differences—but let us also direct attention to our common interests and to the means by which those differences can be resolved.” Or at least, “while defending our own vital inte...

McNamara called the new-old doctrine “Assured Destruction.” (A few years later, a pro-counterforce defense analyst named Donald Brennan would dub this concept “Mutual Assured Destruction” so that it formed the acronym “MAD,” a deprecating term that would soon be adopted by the doctrine’s advocates as well.) To give the term a scientific tinge, McNamara’s chief whiz kid, Alain Enthoven, who’d run the secretary’s “systems analysis” shop since the start of his tenure in the Pentagon, calculated that the Soviets would be sufficiently deterred if the U.S. retaliatory strike could kill 30 percent of their population and destroy half of their industrial capacity.

McNamara even admitted, in a Top Secret memorandum to the president, that he was calculating “the destructive capacity of our force on the hypothetical assumption all of it is targeted on their cities, even though in fact we would not use our forces in that manner if deterrence failed.” This fact—this duplicity—would not change in the subsequent four years of McNamara’s tenure in the Pentagon.

McNamara also took the bolder step of scaling back the Minuteman ICBM program. In his previous three budget memos, he’d wavered between recommending 1,200 to 1,400 missiles by the end of the decade. Now he proposed keeping the total at 1,000. The Chiefs were apoplectic. They’d always distrusted McNamara; now they saw him as a saboteur. LeMay sometimes asked his colleagues, “Would things be much worse if Khrushchev were secretary of defense?” McNamara was touting a “doctrine” based entirely on threatening to launch U.S. nuclear weapons at Soviet citizens and industrial plants, knowing full well that the actual war plan called for aiming almost all of them at military targets—yet he wasn’t giving the Chiefs enough weapons to accomplish the real mission.

Of the three types of strategic nuclear launchers—bombers, submarine-launched ballistic missiles, and land-based ICBMs—the ICBMs were at once the most potent and the most vulnerable: that is, their accuracy made them most capable of destroying blast-hardened targets, such as ICBM silos. Yet, by the same token, their fixed positions made them vulnerable to an adversary’s accurately guided missiles. MIRVs compounded this precariousness. If both sides had only single-warhead ICBMs, an arms race would be senseless: one side could easily match the other’s buildup. But if one side had MIRVs, adding one more missile would mean adding three to ten more warheads—and the ability to destroy, theoretically, three to ten of the other side’s missiles. If both sides had MIRVs, they would both have a dramatic edge—and they would both be vulnerable to the other side’s dramatic edge. In a crisis, a desperate or risk-prone leader might be tempted to launch a first strike, if just to preempt the other side’s leader from launching a first strike. In short, the very existence of MIRVs created a hair-trigger situation, in which both sides might feel an incentive to strike first.

By the time Nixon entered the White House, the SIOP had been modified, a little bit, to allow for three “attack options”: Alpha, which would fire a portion of the arsenal at Soviet and Chinese nuclear weapons sites; Bravo, which covered other military facilities and political-control centers located outside of cities; and Charlie, which would destroy all other military sites, regardless of their location, as well as 70 percent of the industrial floor space inside Soviet and Chinese cities. There was also a sub-option to exclude Moscow and Beijing from Alpha and Bravo strikes. Otherwise, the nuclear war plan offered the president no other choices, and the smallest of those choices—Alpha with the Moscow-Beijing exemption—would fire 1,750 nuclear weapons: hardly a “limited” attack.

Kissinger, an active participant at academic conferences and seminars on nuclear war, wrote, in one volume of essays on the subject, that deterrence “is as much a psychological as a military problem” and that a “threat meant as a bluff but taken seriously is more useful for purposes of deterrence than a ‘genuine’ threat interpreted as a bluff.” In some games of “threats and counter-threats,” he wrote, a “premium will be placed on irresponsibility,” and both sides—Americans and Soviets—will “have to be ready to act like madmen.”

In December 1968, the group produced a RAND monograph titled Rationale for NU-OPTS. It was written by a RAND analyst named James Schlesinger and supervised by Richard Yudkin, an unusual Air Force general—Jewish, bookish, with no experience as any sort of airplane pilot, having earned his three stars entirely as a military strategist. They derived their key insights from Thomas Schelling, a former RAND eminence who had written two books—The Strategy of Conflict and Arms and Influence—which applied mathematical Game Theory to modern warfare, depicting combatants as rational actors pursuing a strategy that maximized their interests, given the strategy that their opponent was most likely to pursue. Schelling imagined scenarios involving small-scale nuclear strikes: shots across the bow to demonstrate resolve, pricks of pain to convey a message to an adversary that more pain would follow unless he halted his aggression. Yudkin and Schlesinger outlined a concept built largely on Schelling’s ideas, suggesting that small-scale nuclear strikes should be built around prepackaged slices of the SIOP—a specific slice for one war scenario, a different slice for another scenario—which they labeled “Selective Attack Options.”

The uncertainties were daunting, the assumptions dubious, slamming into the same analytical obstacles that plagued the architects of the first-strike plan during the Berlin crisis or, for that matter, all the real-life musings on the theory of counterforce: Would the Soviets read the signals correctly? Could they respond in kind, if they wanted to—and would they want to? What if they retaliated with a larger nuclear attack, or with no nuclear attack at all? What would the next move be? How does a president or a commander end the war on favorable terms?

The plan called for launching nearly 200 nuclear weapons at military targets—air bases, bivouacs, and so forth—in the southern region of the USSR, near the Iranian border. Kissinger exploded. “Are you out of your minds?” he screamed. “This is a limited option?” He told them to go back and come up with a smaller plan. A few weeks later, the generals returned. There were two roads leading into Iran from the USSR. The revised plan called for exploding an atomic demolition mine on one of the roads and dropping two nuclear bombs on the other. Kissinger’s eyes rolled. “What kind of nuclear attack is this?” he scoffed. The president of the United States takes the horrible risk of firing nuclear arms in anger for the first time since the end of the Second World War—and he uses only three weapons? Kissinger later told Aaron and Lodal that if the president carried out this plan, Brezhnev would think he was “chicken.” It was the dilemma faced by every policymaker trying to make sense of nuclear weapons and Cold War politics: how to plan a nuclear attack that was large enough to terrify the enemy but small enough to be recognized unambiguously as a limited strike, so that, if the enemy retaliated, he’d keep his strike limited too.

The Joint Chiefs ran periodic simulations of a Soviet nuclear attack, with assistant secretaries and other aides role-playing their bosses. In the first two of these exercises during his term, Carter played himself, an act that both impressed and unnerved the officers in the room; he soon after learned that he was the first president ever to take part in the exercise, a fact that astonished him.

Schmidt’s speech prompted a series of interagency meetings in the Situation Room in the West Wing of the White House. In the months before the speech, several defense analysts had been pushing for precisely the kind of new weapon that Schmidt was requesting. Most of these analysts were members of the European-American Workshop, a low-key group chaired by Albert Wohlstetter, a charismatic consultant who had written some influential studies in the 1950s at the RAND Corporation. The author of Schmidt’s speech, Walter Stützle, director of the German Defense Ministry’s planning office, was an active member of the workshop. Other members had contacts inside the American defense bureaucracy. By the time the interagency review began, a doctrinal rationale for the new missile had come into being. Brzezinski hadn’t given the issue much thought, but his deputy, David Aaron, who’d held a similar post under Kissinger, was an ardent advocate. The SS-20s, Aaron said at one meeting, were creating a “gap in the escalation spectrum”—an odd phrase that was circulating in NATO circles as a result of the workshop’s influence. The premise behind the concept was that the United States needed to match Soviet deployments at every step along the “escalation ladder,” to invoke another metaphor.

Six months before the resolution of the NATO crisis, President Carter and Soviet president Leonid Brezhnev signed the Strategic Arms Limitation Treaty—SALT II, as it was known—the culmination of seven years of negotiations, through the previous two presidents, and the most ambitious nuclear arms control treaty to date. Yet to obtain the Joint Chiefs of Staff’s endorsement, which was necessary to win the Senate’s ratification, Carter had to fund what would be the biggest, most destabilizing missile in the U.S. arsenal. This political tradeoff—an arms buildup in exchange for an arms treaty—had become standard practice by this time, a built-in feature of the arms control process that the Chiefs had honed to an art. It began under Nixon. At the same time that he and Kissinger were unfurling nuclear threats, they were also crafting a diplomacy of “détente,” a complex of forums and agreements to reduce tensions and build trust with the Kremlin—and the Strategic Arms Limitation Talks formed the centerpiece of this effort.

If the Soviets invaded Western Europe, a better use of nuclear weapons, Odom argued, would be to deal with the direct threat—to destroy the Red Army on the battlefield—and this new digital technology presented a chance to do that. Odom took part in the Nuclear Targeting Policy Review with an intense verve. One of his ideas was to scuttle the SIOP, with its rigid preplanned attack options, and to turn SAC into an improvising organization, adapting to scenarios as they developed in a real battle, in real time, and using nuclear weapons in the same way that the Army used artillery or the Air Force amassed tactical fighters for air support of troops on the ground. Odom’s ideas attracted interest but no following. Other military officers, especially those with SAC experience, explained that nuclear planning was a complex business: bomber runs had to be carefully timed, an attack’s expected damage and collateral damage couldn’t be calculated so easily, missiles couldn’t be retargeted on a moment’s notice—the process took days (a few years earlier, it had taken weeks).

Reagan entered the White House with an entourage bent not merely on deterring and containing the Soviet Union but on weakening and rolling back its empire. Thirty-two members of his administration, including Reagan himself, were members of Paul Nitze’s Committee on the Present Danger. In his opening months in office, Secretary of Defense Caspar Weinberger boosted military spending by 13 percent. He revived every nuclear weapons program that Carter had killed (notably the B-1 bomber) and accelerated all the others.

Nitze’s committee had done much to propagate the idea during Carter’s years in the White House, and Reagan named Nitze to be his chief arms negotiator. The chief Soviet affairs specialist on the National Security Council now was Richard Pipes, a Russian history professor at Harvard who had recently written a widely read essay in Commentary magazine called “Why the Soviet Union Thinks It Could Fight and Win a Nuclear War,” which implied that the United States should start thinking along the same lines. A young scholar named Keith Payne, who had coauthored an article in Foreign Policy called “Victory Is Possible” (referring explicitly to victory in a nuclear war), was now a consultant for the Arms Control and Disarmament Agency, where Eugene Rostow, Walt Rostow’s brother, a longtime friend of Nitze’s and cofounder of the Committee on the Present Danger, was director. Richard Perle, the combative defense policy aide to Henry “Scoop” Jackson, the Senate’s leading anti-détente Democrat, was now the assistant secretary of defense for global strategic policy.

Sitting in the White House Situation Room, Reagan heard the briefers from the Joint Staff calmly tell him that, in a Soviet attack, about 5,700 nuclear weapons would explode on American soil, destroying three quarters of the U.S. strategic nuclear forces and killing 80 million Americans. The briefers spelled out the Major Attack Options and Selective Attack Options from PD-59, as well as an option for Launch Under Attack: firing the missiles, especially the land-based ICBMs, before they were hit by the incoming Soviet warheads. (Most of SAC’s officers had figured, from the time of the first centralized war plan, that this is what any president would want them to do.)

When Robert McNamara was secretary of defense, he and his whiz kids analyzed the ABM systems on the drawing board and found them wanting: the interceptors had failed their tests or, to the extent they seemed to pass them, it was because the tests were rigged. A few years into his tenure, McNamara came to a more profound realization: even if the ABM system worked flawlessly, the Soviets could overwhelm it by building—and launching—a few extra missiles, and they could keep building more offensive missiles, more quickly and cheaply than the U.S. could keep building defensive interceptors. At best, then, ABMs would spawn an offense-defense arms race, which the offense—the Soviets—would win. It was this realization that impelled Nixon and Kissinger to negotiate the ABM Treaty with the Soviet Union. McNamara first proposed limits on defensive as well as offensive arms at a 1967 summit in Glassboro, New Jersey, with Soviet prime minister Alexei Kosygin. Initially Kosygin was puzzled, finding the notion of limiting defensive weapons as not only absurd but “immoral.” Over time, though, he and others on the Politburo came to grasp McNamara’s logic and, by the time Nixon entered office, consented to pursuing a treaty.

None of these incidents prompted a presidential crisis, in part because the international climate was calm; few believed that the Soviets had really launched an attack. But the Soviet Union’s false warning on September 26 came at a time of intense nervousness. As it happened, the chief air defense officer on duty, Lieutenant Colonel Stanislav Petrov, figured that the radars had to be mistaken and decided, entirely on his own judgment, not to notify his superiors. If Petrov had been less sure of himself and sent an emergency message alert to the next level—in other words, if he’d been more like his comrades who shot down KAL Flight 007 a month earlier under similar circumstances—World War III might have started that day.

Under Reagan, these sorts of programs were stepped up. In August and September of 1981, an armada of eighty-three U.S., British, Canadian, and Norwegian ships sailed near Soviet waters, undetected. In April 1983, forty U.S. warships, including three aircraft carriers, approached Kamchatka Peninsula, off the USSR’s eastern coast. As part of the operation, Navy combat planes simulated a bombing run over a military site twenty miles inside Soviet territory. The ships and the planes maintained radio silence, jammed Soviet radar, and transmitted false signals; as a result, they avoided detection, even by a new Soviet early-warning satellite orbiting directly overhead. An internal NSA history noted, “These actions were calculated to induce paranoia, and they did.” The simulated bombing run, in particular, incited Yuri Andropov—the former KGB director who had ascended to the Kremlin’s top post after Leonid Brezhnev died toward the end of 1982—to issue an order authorizing air defense units to “shoot to kill” any foreign aircraft intruding into Soviet airspace. Though neither Reagan nor any other senior U.S. official made the connection at the time, this order set the stage for the shoot-down of KAL 007 four months later. In fact, just hours before the Boeing passenger plane mistakenly flew across the border, an American spy plane, sporting a similar radar profile, had been spotted in the same area. (Of course, even if Flight 007 had been a spy plane, shooting it down would have b...