A pilot prospective cohort study was conducted to determine delirium frequency and outcome in cancer patients consecutively hospitalized for terminal care (median stay: 12 days). Screening for delirium symptoms was performed daily, using the Confusion Rating Scale. Patients positive on screening had a diagnostic assessment within 24 hours using the Confusion Assessment Method. Monitoring of symptoms was continued until death. Eighteen (20%) of the 89 study patients were positive on screening at admission. Among the 71 patients free of delirium at admission, the incidence of confirmed delirium was 32. 8% (95% CI, 21.3-44.3%). Patients positive on screening received a higher mean equivalent parenteral daily dose of morphine than other patients (72 mg vs. 41 mg, p = 0.08). Significant symptom improvement occurred in 16 (50%) of the 32 delirious cases. Delirium is a serious and frequent complication in terminal cancer whose outcome may not be as poor as previously considered.