OBJECTIVES--To investigate the incidence of tardive dyskinesia in elderly individuals beginning treatment with antipsychotic drugs and to identify risk factors for the development of tardive dyskinesia in the elderly. DESIGN--A cohort of previously neuroleptic-naive patients was identified at the time of initiation of antipsychotic drug treatment. Patients were evaluated at baseline and followed up at 3-month intervals for periods ranging from 3 to 119 weeks. SETTING--Subjects were recruited from the psychiatric and geriatric medical services of two medical centers, a geriatric institute, and three nursing homes in the metropolitan area of New York City, NY. PATIENTS--Two hundred fifteen individuals older than 55 years have entered the study thus far. Preliminary data are presented for 160 patients who were followed up for at least 1 month. Their age range was 57 to 99 years (mean, 77 years). Seventy-two percent were women and 87% were white. Sixty-seven percent of patients had a diagnosis of organic mental syndrome and 42% had a psychiatric diagnosis. INTERVENTIONS--A naturalistic, longitudinal, repeated assessment paradigm was employed. Assessments included abnormal involuntary movements, extrapyramidal signs, psychiatric symptoms, and medical and drug treatment histories. MAIN OUTCOME MEASURE--The incidence of tardive dyskinesia was determined using a standardized rating instrument and survival analysis. RESULTS--The incidence of neuroleptic-induced dyskinesia was 31% (95% confidence interval, 20%, 42%) after 43 weeks of cumulative neuroleptic treatment. Psychiatric (as opposed to organic) diagnosis and presence of extrapyramidal signs early in treatment were associated with increased tardive dyskinesia vulnerability.