Experimental work has established that a sexual process can occur in African trypanosomes (Jenni, Marti, Schweizer, Betschart, Le Page, Wells, Tait, Paindavoine, Pays & Steinert, 1986; Paindavoine, Zampetti-Bosseler, Pays, Schweizer, Guyaux, Jenni & Steinert, 1986; Tait, personal communication). However, the role of the process in natural populations of trypanosomes is poorly understood. This paper considers what information can be gained from analyses of isoenzyme polymorphism. A cladistic approach is used to help determine whether trypanosome diversity could have been produced by mutation alone. When applied to three East African populations of Trypanosoma brucei it provides evidence that some diversity has arisen through a sexual process; this explains the variation observed within a locality and can account for the evolution of differences between localities. However, the extent to which genetic exchange currently operates is less clear. Analysis of genotype frequencies indicates that agreements with Hardy-Weinberg expectations can be obtained even if genetic exchange exerted no influence over genotype frequencies. Moreover, analysis of joint locus frequencies reveals disequilibrium between loci and that trypanosome populations may be lacking several genotype combinations. Thus, genetic exchange may not occur sufficiently frequently, or in such a way as to break up associations between loci. The relevance of these observations to the evolution of strain differences within T. brucei is discussed.