Elderly psychiatric inpatients were studied to find the prevalence of osteomalacia in those taking anticonvulsant medication (n = 19) and a comparison group not taking these drugs (n = 37). Only one biopsy-proven case was discovered among the anticonvulsant group, and none in the comparison group. This was an unexpectedly low rate. The clinical and biochemical data (including alkaline and acid phosphatase isoenzymes) were further analysed to detect subclinical osteomalacia, but none was found. A difference between the two subgroups was found for total and liver alkaline phosphatase only. The possible explanations for these unexpected findings are discussed. The effect of anticonvulsants on vitamin D and bone metabolism is reviewed and the hypothesis that drug-induced changes in vitamin D metabolism are responsible for the bone changes described in earlier series is questioned.