The objectives of this randomized clinical trial were to investigate the impact of the discontinuation of long-term antipsychotics in residents with dementia in chronic care institutions and to identify clinical predictors of safe discontinuation. Subjects included 34 residents with dementia who were on antipsychotics for more than 6 months and whose behavior was currently stable. Subjects were randomized to either continue receiving their regular dosage of antipsychotics or to receive placebo for 6 months. Early withdrawal from the study was not statistically different between the groups (relative risk [RR] = 1.57, 95% confidence interval [CI] 0.76–3.26), and though not significantly different, subjects in the placebo group were more likely to be withdrawn from the study because of worsening behavior (RR = 1.25, 95% CI 0.33–4.76). Three subjects in the placebo group were withdrawn from the study due to worsening of extrapyramidal symptoms. The active treatment group had more behavioral problems (e.g., physical aggression towards others, p < .05) compared to the placebo group. The placebo group developed more apathy, but balancing this outcome was a relative improvement in congitive functioning. Baseline antipsychotic dose was predictive of behavioral worsening upon discontinuation of long-term antipsychotic drugs. The primary limitation of the study was the small sample size. In conclusion, a trial of discontinuation of antipsychotics should be considered in this population.