The ability to maintain small populations in quasi-natural settings is an issue of considerable importance in biodiversity conservation. The genetic structure of urban common frog (Rana temporaria) populations was determined by allozyme electrophoresis and used to evaluate the effects of restricted intersite migration. Despite the lack of any absolute barrier to movement between ponds, substantial genetic differentiation was found between sites separated by an average of only 2.3 km. Genetic distances between these town ponds correlated positively with geographical distances and were almost twice as great as those found between rural sites separated by an average of 41 km. Measures of genetic diversity and fitness were always lowest in the town, where the degree of subpopulation differentiation (FST =0.388) was high. Population decline was not evident in the town, but molecular and fitness data indicated the presence of genetic drift and inbreeding depression. The long-term survival of artificially restricted populations, particularly of relatively sedentary species, may require molecular monitoring, if genetic diversity is not to be lost by chance when facets of the species niche prove to be poorly understood.