Phenytoin has been shown to induce serum IgA deficiency in patients with epilepsy. We have followed 37 phenytoin-treated patients with reduced serum IgA concentrations for 2-7 years. All anti-epileptic drug treatment was withdrawn in 8 patients. Phenytoin was substituted by other anti-epileptic drugs in 13 patients. Sixteen patients received phenytoin unchanged. The mean serum IgA concentration increased from 0.14 g/l to 0.71 g/l in the patients off drug treatment, and from 0.20 g/l to 0.84 g/l in the patients changing to other drugs. The mean IgA concentration increased from 0.21 g/l to 0.37 g/l in the patients with phenytoin unchanged, due to the increase of the IgA levels in a few patients who initially showed a slightly depressed IgA concentration. The overall increase of IgA concentrations was significant. When tested separately in each patient group, the difference was significant for the patients off all drugs, and for those changing from phenytoin to other drugs. The IgM concentration increased significantly when all patients were considered together. This was due to the increase of IgM in the patients no longer receiving phenytoin therapy. Withdrawal of phenytoin did not influence the IgG concentrations.