Risperidone has been used to treat behavioral symptoms, such as delusions and agitation, in people with Alzheimer's disease. The relationship between magnitude and variability of risperidone and 9-hydroxy risperidone exposure and the relationship with time to discontinuation of the medication were explored. Sixty-five subjects from the Clinical Antipsychotic Trial of Intervention Effectiveness-Alzheimer's Disease Trial that received risperidone were included in this study. Eighteen subjects completed the study without switching medication (completers on risperidone), whereas 47 discontinued the medication. Those who discontinued were divided into 2 groups according to responsiveness to therapy. Using Cox proportional survival regression analysis, we estimated time to discontinuation and factors associated with treatment discontinuation including age, dose, body mass index, neuropsychiatric inventory baseline score, and average exposure (area under the curve [AUC]) to risperidone and 9-hydroxy risperidone. Twenty-four and 17 subjects discontinued therapy because of inadequate therapeutic effect and side effects, respectively (6 subjects were excluded because of missing information about reason for switching or discontinuation). Discontinuation hazards for those with a higher than median AUC of the metabolite were 2.54 (P = 0.029; inadequate and side effect group combined) and 3.48 (P = 0.025; inadequate effect group) times that of those in the lower than median AUC group. None of the other covariates contributed significantly to the switching hazard. Risperidone metabolite, 9-hydroxy risperidone concentrations, correlated with the risk of switching or discontinuing the medication, suggesting that 9-hydroxy risperidone contributes to adverse events and intolerability in dementia patients.