Some statistical analysis and experiments have been carried out on the single-locus VNTR reference databases collected separately from the Chinese in Hong Kong and Singapore to assess the issues of sample size and representativeness in the context of forensic science. We employed the bootstrap method and the Pearson chi-squared test for independence, and found that the different reference databases have very little effect in assessing the frequencies of occurrence of variable number of tandem repeats (VNTR) profiles. The results of random match probability assessment also indicate that the sample sizes studied appeared adequate to provide representative data. It has been found that a battery of four loci was enough to distinguish the 458 individuals under study.