Random amplified polymorphic DNA (RAPD) markers are sensitive to changes in reaction conditions and may express polymorphisms of nongenetic origin. Taxa with variable chromosome numbers are particularly challenging cases, as differences in DNA content may also influence marker reproducibility. We addressed these problems by comparing RAPD and amplified fragment length polymorphism (AFLP) analyses of clonal identity and relationships in a chromosomally variable arctic plant, the polyploid Saxifraga cernua, which has been thought to be monoclonal over large geographical distances. Fifty-seven plants from four Greenland populations were analysed using a conservative scoring approach. In total, 26 AFLP and 32 RAPD multilocus phenotypes (putative clones) were identified, of which 21 were identical and each of the remaining five AFLP clones was split into two to three very similar RAPD clones. This minor difference can be explained by sampling error and stochastic variation. The pattern observed in Greenland corroborates our previous results from Svalbard, suggesting that rare sexual events in S. cernua are sufficient to maintain high levels of clonal diversity even at small spatial scales. We conclude that although AFLP analysis is superior in terms of efficiency, RAPDs may still be used as reliable markers in small low-tech laboratories.