鈥� Twenty-one hospitalized patients over 50 years of age who had tardive dyskinesia were studied for 13 months. After withdrawal of neuroleptics and antidepressants for three months, dyskinetic symptoms abated in 12 patients and persisted in nine. Discriminant function analysis showed that the persistent and reversible dyskinesia groups could be clearly separated and that the best discriminator was the number of drug-free intervals. The persistent dyskinesia group had had significantly longer neuroleptic treatment (mean, 10.8 yr) and a greater number (mean, 5.6) of drug interruptions of at least two months' duration each than did the reversible dyskinesia group. Our finding, as well as the literature reviewed, does not support the commonly held notion that frequent lengthy interruptions of long-term drug treatment reduce the incidence of persistent dyskinesia, at least in patients who are otherwise predisposed to the development of tardive dyskinesia.