Polysomnographic studies of Parkinson's disease (PD) patients with visual hallucinations (VH) usually reveal short, fragmented rapid eye movement (REM) sleep, with lower sleep efficiency and reduced total REM sleep. Quetiapine has been demonstrated in open-label trials to be effective for the treatment of insomnia and VH in PD. To confirm quetiapine's efficacy in improving VH, and to determine whether the mechanism was due to its effect on REM sleep architecture, we performed a pilot, double-blind, placebo-controlled study. Sixteen PD patients experiencing VH were recruited. Eight patients were randomized to quetiapine and eight patients to placebo. Patients underwent pre- and post-treatment polysomnography. The Clinical Global Impression Scale (CGIS), Brief Psychiatric Rating Scale (BPRS), and Unified Parkinson Disease Rating Scale (UPDRS) motor subscale were obtained. There were no differences in baseline characteristics between the treatment arms except that the placebo group had more sleep in stage REM (74.7 min vs. 40.1 min; p < .001). Data were imputed for all patients who prematurely discontinued (four quetiapine and one placebo) in an intention-to-treat analysis. The average quetiapine dose was 58.3 mg/day. While there was no significant difference in the change in REM duration pre- vs. post-treatment in either arm, patients randomized to quetiapine improved on the CGIS (p = .03) and the hallucination item of the BPRS (p = .02). No difference was noted in the UPDRS motor scores. Despite the small sample, this is the first double-blind trial to show quetiapine's efficacy over placebo in controlling VH in the PD population. However, normalization of sleep architecture was not supported as the mechanism.