This report describes a suite of benchmarks for Prolog implementation research. It includes an explanation of the format of the suite, which is meant to facilitate use of the benchmarks. The principal idea of this format is to maintain for each benchmark a master file from which particular instances -- for particular Prolog execution systems, for particular statistics to capture, etc. -- are generated automatically using a preprocessor. A preprocessor provided with the suite for this purpose is described, along with a related utility and a simple framework for execution time measurement. Source code for these is appended. Possibilities for future work with respect both to this suite and to Prolog benchmarking more generally are discussed briefly. For each benchmark in the suite, source code and execution times under C Prolog and Quintus Prolog (compiled) on a Sun 3/60 are appended.