Guidance exists on how to count matches between samples in a crime sample database but we are unable to locate a definition of how to estimate a match rate. We propose a method that does not proceed from the match counting definition but which has a strong logic.