We have critically evaluated 3 techniques for the detection of mutations conferring drug resistance of Plasmodium falciparum, using samples containing known numbers of well-characterized parasites and artificial mixtures of these parasites at known proportions. We compared the sensitivity and specificity of mutation-specific polymerase chain reaction (MS-PCR), polymerase chain reaction followed by restriction enzyme digestion at polymorphic sites (PCR/RFLP), and a dot-blot/probe hybridization technique, for detection of point mutations at nucleotide 323 of the P. falciparum dihydrofolate reductase gene (dhfr) that confer resistance to the antimalarial drug pyrimethamine. We have also investigated the benefits in terms of sensitivity and reproducibility of the incorporation of radiolabelled nucleotides into the PCR/RFLP assay. We found that MS-PCR was very sensitive--at least 10 parasites could be detected in a sample--but non-specific amplification resulted in erroneous typing of some samples. PCR/RFLP was less sensitive; 10 parasites per sample could not always be detected, but the technique was specific. The addition of radiolabelled nucleotides to the assay did not markedly improve the sensitivity but the results were easier to read and there was less subjectivity in scoring the results. The dot-blot/probe hybridization technique was specific and sensitive, with similar levels of specificity and sensitivity to PCR/RFLP. On balance, the dot-blot/probe hybridization technique seems best suited to large-scale epidemiological surveys of genes associated with antimalarial drug resistance.