The sugarcane varieties currently grown commercially in South Africa arose from early crosses between varieties of Saccharum officinarum, S. spontaneum and S. barberi Subsequently, there were approximately six generations of successive in-breeding between these hybrids to produce the current varieties. In the present investigation RAPDs were used to determine the genetic variation in some elite members of the germplasm collection. DNA was extracted from leaf rolls from three stools of each variety. Three hundred and fifty six loci were amplified with 41 random 10-mer primers (Operon Technologies), i.e. an average of 8.7 fragments per primer. From these data, 160 loci, obtained with 15 primers, were scored for presence/absence across the 20 sugarcane varieties and the data analysed with NTSYS-pc. These primers were used as they produced the most clear and unambiguous banding profiles with at least a few scorable polymorphisms. Little genetic variation was observed between the commercial varieties and those at advanced stages of selection. The largest difference was between CB38-22 and N16 (80% similarity) while no differences were detected between 79M955 and 8OL432. The genetic distances between S. spontaneum (Coimbatore), S. officinarum (Black Cheribon), hybrids of these from early in the sugarcane genealogy, and the commercial varieties were much greater (65%). In contrast to the low variation detected with RAPDs numerous polymorphic bands were resolved using the tetranucleotide repeat (GATA)2(GACA)2 and three telomere primer sequences. The variation detected with these primers between die clones ranged from 15% to 74%.