This report describes the evaluation and treatment of delirium in the cancer patient in a major comprehensive cancer center. Ninety consecutive cases of delirium seen by the inpatient psychiatry consultation/liaison service were analyzed in a retrospective fashion to evaluate demographic information, alcohol use, central nervous system disease, coexisting medical disease, and past psychiatric history. Delirium cases were divided into hyperalert, hypoalert, and mixed subtypes. For these three subtypes, medication profiles including dose of medication, duration of delirium, outcome, and the venue where the delirium began were also evaluated. The hyperalert subtype of delirium was the commonest type observed (71%) and had the shortest duration (P < 0.0001) and best outcome (P < 0.001). The patients with a hyperalert delirium subtype were treated with the least amount of haloperidol (P < 0.0001). Patients were delirious for longer when the delirium began in the intensive-care units (P < 0.04). In general, patients who received no haloperidol experienced delirium of longer duration (P < 0.02) than those receiving haloperidol. Since the data represent patients who were referred for psychiatric treatment, this may explain the increased number of hyperalert deliriums and, therefore, the generalizability of the results is limited. Delirium in the cancer patient is particularly problematic given the coexisting medical problems these patients experience. Because the outcome of delirium is better when the duration is shorter, it is important for clinicians to be sensitive to early symptoms so that treatment can be implemented faster, leading to less morbidity and mortality.