There is considerable debate regarding the effect of population subdivision (heterogeneity) on the probability of a chance or coincidental match between two DNA samples studied with respect to multiple, polymorphic genetic markers. We have theoretically investigated the relationship between the average similarity between two randomly chosen DNA samples and the probability of an identical match between these samples, and population subdivision. Our results demonstrate that the average similarity and the match probability is smaller when population heterogeneity exists as compared to a random mating population with identical gene frequencies, for realistic values of heterogeneity. In other words, ignoring subdivision provides numerical values only slightly larger than the true values and are, thus, conservative.