Peptide-Mass Fingerprinting (PMF) encompasses a number of techniques for protein characterization which have as a first step the cleaving of target proteins by chemical or enzymatic reagents. Software systems exist which perform similar analyses. However, this is the first study which examines theoretically the effectiveness of the particular reagents for PMF. In this study, the task of PMF was to identify every sequence in a non-redundant protein database, via "in silico" digestion with theoretical proteases. From these experiments, some conclusions are drawn about the characteristics of better reagents and the experimental conditions which are more likely to be useful for PMF. The need for strongly non-redundant databases is also highlighted.