In a previous paper, isolation by distance at RFLP loci was demonstrated in 10 populations of sea beet growing on the Dorset (U.K.) coast. In the same populations there was no evidence of isolation by distance at isozyme loci. Here, the data are reanalysed to reduce biases resulting from nonrandom mating within the populations; the contrast between RFLPs and isozymes remains very clear. Separate analysis of cliff-top populations shows strong isolation by distance at both RFLP and isozyme loci, whereas in populations on sea walls around the perimeter of Poole Harbour there is isolation by distance at RFLP loci but not isozyme loci. Thus, whatever the cause of the difference between RFLP and isozyme loci, it occurs in the harbour populations only. There was no evidence that founder effects are a significant source of variation between populations on the perimeter of the harbour. This is very different from a previous finding of strong founder effects between populations on Furzey Island (in Poole Harbour).