The issue of whether genetic exchange occurs at a significant frequency in natural populations of Trypanosoma brucei is controversial and one of the arguments against a high frequency has been the apparent lack of host infections with mixtures of trypanosome genotypes. Three minisatellite markers (MS42, CRAM, 292) within the coding regions of three genes have been identified and PCR based methods developed for detecting variation at these loci using crude lysates of infected blood as templates. Initial PCR analysis, using primers flanking the repeats, of DNA from two cloned stocks of the parasite has shown that two DNA fragments of different size were amplified from each stock. Analysis of the inheritance of these fragments into the F1 progeny of crosses demonstrated that the different size fragments were alleles that segregated in a Mendelian manner. The alleles at each of the three loci segregated independently consistent with their localisation on three different chromosomes. Analysis of a series of cloned isolates from tsetse flies showed that these loci were highly variable giving heterozygosities of 94% and the identification of 12 distinct alleles in a sample of 17 cloned isolates. In order to determine whether isolates are heterogeneous in terms of trypanosome genotype, the allelic variation at these three loci was examined in uncloned samples from tsetse flies isolated in Kiboko, Kenya and Lugala, Uganda. A significant proportion of the isolates (36% in Lugala and 47% in Kiboko) contained more than two alleles at one or more of the loci thus demonstrating that a high proportion of tsetse flies were infected with more than one genotype of trypanosomes. This was established, unequivocally, for two isolates by generating a series of cloned trypanosome lines from each and determining the genotype of each clone; one isolate (927) contained seven different genotypes with a high proportion of the possible combinations of alleles at each locus. These results indicate the possibility of frequent genetic exchange in the field, they imply that a significant proportion of mammalian hosts must contain mixtures of different trypanosome genotypes and they demonstrate the advantages of using minisatellite markers for the analysis of the population structure of T. brucei.