A common opinion nowadays, I suspect, is that Prolog is a neat hack that ran wildly out of control. And it is an opinion that is easy defend, and one with which I even have a lot of sympathy: not only does Prolog have substantial and not-really fixable problems as a 'serious' programming language, but it was also, in the aftermath of the 5th-generation hype, the inspiration for a lot of embarrassingly bad theoretical and quasi-theoretical research on 'logic' programming in the late 1980's and early 1990s. On the other hand, Prolog is also distinguished by some of the best books on progamming I have ever read: not just O'Keefe's 'The Craft of Prolog', but also, e.g., Sterling and Shapiro's 'Art of Prolog' crowd into the (depressingly small) queue formed behind the likes of 'Structure and Interpretation', 'the Science of Programming' and 'Programming Tools'. The existence of such books means that Prolog must have gotten _something_ substantial right.
Further, while in theory I divide the the set of all programming languages into clean Lisp dialects (i.e. scheme, ml, haskell) on the one hand, and other programming languages that are inadequate to the extent that they diverge from the Scheme/ML model on the other, I find that a lot of the time it is actually Prolog that provides the best tool for modelling the transaction-handling systems that I have to deal with in the course of earning my bread.
Whether you use Prolog or not, if you are serious about programming then you want to have a copy of this, simply because it shows how a world class programmer negotiates an unusual, but interesting, programming paradigm. And, as O'Keefe himself is, or at least used to be, fond of pointing out, your skill as a programmer is substantially correlated with the number of different such paradigms that you understand properly, and not very much with anything else.
Highly recommended if you are really interested in advanced programming.