Reversible tardive dyskinesia (TD) outcomes have been reported in long-term neuroleptic (NL)-treated patients. In this study the course of TD outcomes was followed-up for 3 years in a population of 125 institutionalized schizophrenic patients (mean age 57.8 years) receiving continuous NL treatment. Tardive dyskinesia occurrence and severity were assessed by means of the Rockland Simpson Scale (RSS). The prevalence of TD rose from 39.2% at the first examination to 52.8% at last follow-up examination; however, 28.6% of TD-affected patients recovered and 30% improved. Significant risk factors for a persistent TD outcome result included age over 56 years, duration of illness over 30 years, and a total RSS score over 22. Cumulative NL exposure, over 3550 g of chlorpromazine equivalents, was also a significant risk factor for TD. Results from this study confirm that there is the possibility of improvement and remission in an aged, long-term institutionalized population of TD patients. In this report we point out prognostic factors for positive outcome.