BACKGROUND: Delirium in patients admitted to the intensive care unit (ICU) is a serious complication potentially increasing morbidity and mortality. The aim of this study was to investigate the impact of fluctuating sedation levels on the incidence of delirium in ICU.\n\nMETHODS: A prospective cohort study of adult patients at three multidisciplinary ICUs. The Richmond Agitation and Sedation Scale (RASS) and the Confusion Assessment Method for the ICU were used at least twice a day.\n\nRESULTS: Delirium was detected at least once in 65% of the patients (n = 640). Delirious patients were significantly older, more critically ill, more often intubated, had longer ICU stays, and had higher ICU mortality than non-delirious patients. The median duration of delirium was 3 days (interquartile range: 1;10), and RASS was less than or equal to 0 (alert and calm) 91% of the time. The odds ratio (OR) for development of delirium if RASS changed more than two levels was 5.19 when adjusted for gender, age, severity of illness, and ICU site and setting. Continuous infusion of midazolam was associated with a decrease in delirium incidence (OR: 0.38; P = 0.002).\n\nCONCLUSIONS: Fluctuations in sedation levels may contribute to development of delirium in ICU patients. The risk of developing delirium might be reduced by maintaining a stable sedation level or by non-sedation.