ContextAntipsychotics are often used in managing symptoms of terminal delirium, but evidence is limited. ObjectivesTo explore the comparative effectiveness of haloperidol with as-needed benzodiazepines (HPD) vs. chlorpromazine (CPZ) vs. levomepromazine (LPZ) for agitated delirium in the last days. MethodsA prospective observational study was conducted in two palliative care units in Japan. Adult cancer patients who developed agitated delirium with a modified Richmond Agitation-Sedation Scale (RASS-PAL) of one or more were included; palliative care specialist physicians determined that the etiology was irreversible; and estimated survival was 3 weeks or less. Patients treated with HPD, CPZ, or LPZ were analyzed. We measured RASS, NuDESC, Agitation Distress Scale (ADS), and Communication Capacity Scale (CCS) on Days 1 and 3. ResultsA total of 277 patients were enrolled, and 214 were analyzed (112 in HPD, 50 in CPZ, and 52 in LPZ). In all groups, the mean RASS-PAL score significantly decreased on Day 3 (1.37 to -1.01, 1.87 to -1.04, 1.79 to -0.62, respectively; P < 0.001); the NuDESC and ADS scores also significantly decreased. The percentages of patients with moderate to severe agitation and those with full communication capacity on Day 3 were not significantly different. The treatments were well-tolerated. While one-fourth of HPD group changed antipsychotics, 88% or more of CPZ and LPZ groups continued the initial antipsychotics. ConclusionHaloperidol with as-needed benzodiazepine, chlorpromazine, or levomepromazine may be effective and safe for terminal agitation. Chlorpromazine and levomepromazine may have an advantage of no need to change medications.