We conducted a systematic examination of the experience of delirium in a sample of 154 hospitalized patients with cancer. Patients all met DSM-IV criteria for delirium and were rated with the Memorial Delirium Assessment Scale as a measure of delirium severity, phenomenology, and resolution. Of the 154 patients assessed, 101 had complete resolution of their delirium and were administered the Delirium Experience Questionnaire (DEQ-a face-valid measure that assesses delirium recall and distress related to the delirium episode). Spouse/caregivers and primary nurses were also administered the DEQ to assess distress related to caring for a delirious patient. Fifty-four (53.5%) patients recalled their delirium experience. Logistic-regression analysis demonstrated that short-term memory impairment (odds ratio [OR] = 38.4), delirium severity (OR = 11.3), and the presence of perceptual disturbances (OR = 6.9) were significant predictors of delirium recall. Mean delirium-related distress levels (on a 0-4 numerical rating scale of the DEQ) were 3.2 for patients who recalled delirium, 3.75 for spouses/caregivers, and 3.09 for nurses. Logistic-regression analysis demonstrated that the presence of delusions (OR = 7.9) was the most significant predictor of patient distress. Patients with "hypoactive" delirium were just as distressed as patients with "hyperactive" delirium. Karnofsky Performance Status (OR = 9.1) was the most significant predictor of spouse/caregiver distress. Delirium severity (OR =5.2) and the presence of perceptual disturbances (OR =3.6) were the most significant predictors of nurse distress. In conclusion, a majority of patients with delirium recall their delirium as highly distressing. Delirium is also a highly distressing experience for spouses/caregivers and nurses who are caring for delirious patients. Prompt recognition and treatment of delirium is critically important to reduce suffering and distress.