Delirium is the most common and distressing neuropsychiatric syndrome in cancer patients. It has a negative effect on symptom assessment, patient-clinician communication, decision making, and survival. Approximately 50% of patients with hyperactive or mixed delirium continue to experience agitation despite low-dose haloperidol, which can be particularly distressing to patients and caregivers. Neuroleptic dose escalation, benzodiazepine rotation, combination therapy, and neuroleptic withdrawal may provide good control of agitation. However, no delirium trials have ever been completed to directly compare these options; such trials are urgently needed to improve the quality of life of patients with this devastating syndrome. Our long-term goal is to develop evidence-based therapy for the palliation of delirium in cancer patients. The proposed study is a high-impact, 4-arm, multi-center, double-blind, double-dummy, randomized controlled trial to compare the effect of neuroleptic dose escalation, benzodiazepine rotation, combination therapy, and neuroleptic withdrawal in the treatment of persistent agitated delirium in cancer patients admitted to acute palliative care units (APCUs). We hypothesize that benzodiazepine rotation and combination therapy are particularly effective against agitated delirium. The primary specific aim of this study is to compare the effect of neuroleptic dose escalation, benzodiazepine rotation, combination therapy, and neuroleptic withdrawal on the change in Richmond Agitation Sedation Scale (RASS) over 24 hours in patients who did not experience a response to low-dose haloperidol. The second aim is to compare the effects of these treatments on (1) rescue medication use, (2) the proportion of patients in the target RASS range, (3) perceived comfort by caregivers and bedside nurses, (4) delirium-related distress in caregivers and nurses, (5) achievement of the proxy comfort goal, (6) symptom expression, (7) delirium severity, (8) adverse effects, and (9) quality of end-of-life care. The third aim is to identify novel predictive markers of response to haloperidol and lorazepam. After obtaining surrogate consent, we will administer the study medications under a titration scheme and monitor the participants closely until discharge. This study is highly innovative because (1) multiple previously untested treatment strategies are studied, (2) the outcome measures (e.g., proxy comfort goals) are novel, (3) biomarkers are included as potential predictors of treatment response, (4) the patient population (i.e., short survival) is unique, and (5) the study setting (i.e., the APCU) is distinctive. Successful completion of this definitive study will identify the optimal strategy to reduce agitation, improve patient comfort, and palliate delirium-related distress; provide timely data to address the ongoing debate regarding the proper use of haloperidol and lorazepam in delirium; inform novel strategies to monitor patients with delirium; and stimulate further studies to identify better strategies to palliate persistent agitation and improve patients' quality of life.