Tardive dyskinesia (TD) rates with second-generation antipsychotics (SGAs) are considered to be low relative to first-generation antipsychotics (FGAs), even in the particularly vulnerable elderly population. However, risk estimates are unavailable for patients naïve to FGAs. Therefore, we aimed to determine the TD incidence in particularly vulnerable, antipsychotic-naïve elderly patients treated with the SGA risperidone or olanzapine. The present work describes a prospective inception cohort study of antipsychotic-naïve elderly patients aged X55 years identified at New York Metropolitan area in-patient and out-patient geriatric psychiatry facilities and nursing homes at the time of risperidone or olanzapine initiation. At baseline, 4 weeks, and at quarterly periods, patients underwent assessments of medical and medication history, abnormal involuntary movements, and extra-pyramidal signs. TD was classified using Schooler-Kane criteria. Included in the analyses were 207 subjects (age: 79.8 years, 70.0% female, 86.5% White), predominantly diagnosed with dementia (58.9%) or a major mood disorder (30.9%), although the principal treatment target was psychosis (78.7%), with (59.4%) or without (19.3%) agitation. With risperidone (n ¼ 159) the cumulative TD rate was 5.3% (95% confidence interval (CI): 0.7, 9.9%) after 1 year (mean dose: 1.0±0.76 mg/day) and 7.2% (CI: 1.4, 12.9%) after 2 years. With olanzapine (n ¼ 48) the cumulative TD rate was 6.7% (CI: 0, 15.6%) after 1 year (mean dose: 4.3±1.9 mg/day) and 11.1% (CI: 0, 23.1%) after 2 years. TD risk was higher in females, African Americans, and patients without past antidepressant treatment or with FGA co-treatment. The TD rates for geriatric patients treated with risperidone and olanzapine were comparable and substantially lower than previously reported for similar patients in direct observation studies using FGAs. This information is relevant for all patients receiving antipsychotics, not just the especially sensitive elderly.