Cepaea nemoralis at a site on Fyfield Down, Wiltshire, England, have been surveyed, using mark, release, recapture, for 23 years. The population has morph frequencies that do not appear to match the background. At the start of the survey there was heavy predation by birds, but this soon ceased and the vegetation in the area became more uniform, probably because of the destruction of rabbits by myxomatosis. Over the study period the population fluctuated in size but morph frequencies remained almost unchanged. Variation in recruitment, rather than survival, is responsible for variation in numbers. The morph frequency of captured snails differs between juveniles and adults. We suggest that this effect is due to differences in the extent to which large and small individuals move in the open (exposure). This in turn is due to changes in heating properties as the animals increase in size. If the largest selection pressure affecting morph frequency arises from visual predation, then the lack of predators, and lack of evidence of change in survival rate are consistent with the invariant morph frequencies. The extent to which studies of the present kind can detect selection is discussed.