Objective: Delirium is a frequent problem in the intensive care unit (ICU) associated with poor prognosis. Delirium in the ICU is underdiagnosed by nursing and medical staff. Several detection methods have been developed for use in ICU patients. The aim of this study was to compare the value of three detection methods (the Confusion Assessment Method for the ICU [CAM-ICU], the Intensive Care Delirium Screening Checklist [ICDSC] and the impression of the ICU physician with the diagnosis of a psychiatrist, neurologist, or geriatrician). Design: Prospective study. Setting and patients: During an 8-month period, 126 patients (mean age 62.4 years, sd 15.0; mean Acute Physiology and Chronic Health Evaluation II score 20.9, sd 7.5) admitted to a 32-bed mixed medical and surgical ICU were studied. Measurements: The included patients were assessed independently by trained ICU nurses using either the CAM-ICU or the ICDSC. Furthermore, the ICU physician was asked whether a patient was delirious or not. A psychiatrist, geriatrician, or neurologist serving as reference rater diagnosed delirium using established criteria. Main results: The CAM-ICU showed superior sensitivity and negative predictive value (64% and 83%) compared with the ICDSC (43% and 75%). The ICDSC showed higher specificity and positive predictive value (95% and 82% vs. 88% and 72%). The sensitivity of the physicians view was only 29%. Conclusions: ICU physicians underdiagnose delirium in the ICU, which underlines the necessity of standard evaluation in all critically ill patients. In our mixed ICU population, the CAM-ICU had a higher sensitivity than the ICDSC.