Genetic diversity of malaria parasites represents a major issue in understanding several aspects of malaria infection and disease. Genotyping of Plasmodium falciparum infections with polymerase chain reaction (PCR)-based methods has therefore been introduced in epidemiological studies. Polymorphic regions of the msp1, msp2 and glurp genes are the most frequently used markers for genotyping, but methods may differ. A multicentre study was therefore conducted to evaluate the comparability of results from different laboratories when the same samples were analysed. Analyses of laboratory-cloned lines revealed high specificity but varying sensitivity. Detection of low-density clones was hampered in multiclonal infections. Analyses of isolates from Tanzania and Papua New Guinea revealed similar positivity rates with the same allelic types identified. The number of alleles detected per isolate, however, varied systematically between the laboratories especially at high parasite densities. When the analyses were repeated within the laboratories, high agreement was found in getting positive or negative results but with a random variation in the number of alleles detected. The msp2 locus appeared to be the most informative single marker for analyses of multiplicity of infection. Genotyping by PCR is a powerful tool for studies on genetic diversity of P. falciparum but this study has revealed limitations in comparing results on multiplicity of infection derived from different laboratories and emphasizes the need for highly standardized laboratory protocols.