The collection of buccal cells provides a noninvasive method for obtaining DNA for genetic studies. Here we report the results on buccal cell genotyping from our ongoing study of childhood leukemia in Northern California. We have collected buccal samples from children ranging in age from 4 months to 15 years using an interviewer- or nurse-administered protocol using a cytology brush. Initial results of the genotyping, including the glutathione S-transferase mu, glutathione S-transferase theta, NAD(P)H:quinone oxidoreductase, and methylenetetrahydrofolate reductase polymorphisms, were disappointing because many specimens contained little DNA, failed repeated attempts at PCR amplification, and produced unreliable results. Here we evaluate a solution to the problem that involves whole genome amplification using the improved primer extension preamplification methodology. Sixty cases of pediatric acute leukemia were studied; five PCR-based genotypes were attempted using buccal cell DNA and whole genome amplified (WGA) buccal DNA. Results were compared with genotyping results using DNA isolated from peripheral whole blood or bone marrow for each child. The standard buccal protocol failed to yield successful PCR reactions in 30-57% of specimens, whereas WGA-buccal was markedly more efficient (2-5% failed PCR). A success rate of 100% was achieved with one repeat test of the failed WGA-PCR reactions. Misclassification of genotype was common for the glutathione S-transferase theta marker using the standard buccal procedure. The WGA-buccal protocol, however, produced genotyping results fully concordant with the referent blood or bone marrow DNA results for all five loci. DNA yields were increased by WGA to allow for approximately 900 PCR reactions/brush. WGA is very useful for improving the efficiency and validity of PCR-based genotyping in pediatric populations.