Prolog and Lisp benchmark timings are compared on the VAX 8600. The measured Lisp-to-Prolog time ratio varies between 0.143 and 2.83. These differences between Prolog and Lisp performance can be explained by studying the structure of the benchmarks and the language implementations. Previous authors have used such measurements as evidence that one language is "better" than the other; their works are summarized. The issues involved in comparing two languages are treated, as well as the assumptions involved in interpreting language performance data.