This comprehensive volume is the product of an intensive collaborative effort among researchers across the United States, Europe and Japan. The result -- a change in the way we think of humans and computers.
So, it's a little out of date (erm, first published in 1986). If you look past the now-archaic technical examples, there is plenty of solid theory, upon which many of today's best practices are based.
Like any collection, some essays are stronger than others. But I'd still recommend it for anyone who wants to know more about the theoretical foundations of HCI.