When the smaller of two peaks at an STR locus is less than 70% the magnitude of the larger peak at that locus, the disparity is typically taken to be an indication that there is more than one contributor of template DNA to the sample being tested. An analysis of 1,763 heterozygous allele pairs suggests that a peak height imbalance threshold that varies with the magnitude of the peaks being evaluated at a locus is superior to a fixed threshold. Identifying samples that are likely to be mixtures and those that are likely to have arisen from a single source is accomplished more reliably when a statistically based, magnitude-dependent peak height imbalance threshold is used. The amelogenin locus was found to behave in a similar fashion and was also found to have no systematic bias that favored the amplification of Y or X alleles.