The preferential PCR amplification of one allele relative to another in a heterozygous sample could result in an incorrect or ambiguous genetic typing of that sample. There are several mechanisms that could potentially lead to such preferential PCR amplification. First, preferential amplification can result from significant GC% differences between alleles if the conditions of the reaction (denaturation temperature (Tden), duration at the Tden' salt and co-solvent concentrations, etc.) allow the denaturation of one allele but not the other (differential denaturation). For example, the DQa1.1, -1.2, and -1.3 alleles of the HLA-DQa locus do not amplify at a Tden < 89 degrees C; these same conditions still allow amplification of the DQa2, -3, and -4 alleles. However, no differences in amplification efficiency were found between the different HLA-DQa alleles when the Tden was set at the recommended Tden of 94 degrees C, even after as many as 102 cycles of amplification. Second, for PCR-based genetic typing systems in which the PCR products from different alleles differ in length, preferential amplification of the shorter allelic product can occur. Experiments in which the variable number tandem repeat (VNTR) marker D17S5 (YNZ22) was amplified under various conditions suggest that the smaller allelic products are amplified preferentially when Taq polymerase is limiting. Preferential amplification of VNTR alleles can also occur if the target DNA is sufficiently degraded. Third, when the initial number of genomes sampled is very small, stochastic fluctuation in the number of copies of each allele can result in what appears to be preferential amplification. Finally, less efficient priming of DNA synthesis of one allele versus another can occur because of mismatches between the primer and the specific allelic template, resulting in preferential amplification of the other allele. General strategies to avoid preferential amplification are discussed.