Dementia-related psychosis (DRP) is common among patients with Alzheimer’s disease (AD), Parkinson’s disease (PD), dementia with Lewy bodies (DLB), fronto-temporal dementia (FTD), and vascular dementia (VaD) and is associated with poor outcomes. HARMONY (NCT03325556) was a phase 3, placebo-controlled, randomized, relapse-prevention study evaluating the efficacy and safety of pimavanserin for treating hallucinations and delusions associated with DRP. Patients with dementia and moderate-severe psychosis received open-label (OL) pimavanserin for 12 weeks. Patients with sustained response (≥30% reduction in scale for the assessment of positive symptoms hallucinations+delusions total score AND clinical global impression-improvement score of much/very much improved) at weeks 8 and 12 were randomized 1:1 to continue pimavanserin or receive placebo for up to 26 weeks in the double-blind (DB) period. The primary endpoint was time from randomization to relapse of DRP. 392 patients enrolled. 217 (61.8%) eligible patients experienced sustained response and were randomized. OL response was similar regardless of dementia subtype (randomization rates: 59.8% AD, 71.2% PDD, 71.4% VaD, 45.5% DLB, 50.0% FTD), baseline disease characteristics, age, dementia severity, or previous drug therapy. The study stopped early for superior efficacy when a prespecified interim analysis revealed >2.8-fold reduction in risk of relapse with pimavanserin (hazard ratio: 0.353; 95% CI: 0.172, 0.727; 1-sided p=0.0023). Adverse event rates were low and balanced (OL: 36.2%; DB: 41.0% pimavanserin, 36.6% placebo). No negative trends for worsening in cognition (mini-mental state examination) or motor function were observed. The HARMONY study demonstrated a robust decrease in hallucinations and delusions and significant maintenance of efficacy with pimavanserin treatment in DRP.