Variation at three microsatellite loci was investigated in five populations of sea beet on the Dorset coast. Genetic variation among the populations was estimated using two different methods. Estimates of F-ST were obtained using the program FSTAT, and estimates of R-ST were obtained using the program AMOVA. These different estimates of genetic distance were used to test for isolation by distance with partial matrix correspondence tests and were compared with patterns of variation in F-ST estimates from isozymes and single-copy nuclear restriction fragment length polymorphisms (RFLPs). F-ST from isozymes, F-ST from RFLPs and R-ST from microsatellites all showed evidence of isolation by distance, whereas F-ST from microsatellites showed strong similarities within populations that were not related to distance. These results suggest that, compared with isozymes and RFLPs, mutation rates at microsatellite loci are high in these populations. They also suggest that stepwise mutation processes are occurring at these loci. The data provide valuable evidence that all three types of marker can detect similar patterns of population structure. The results are discussed with respect to estimates of genetic distance at microsatellite loci in human populations.