OBJECTIVE Although fever is common in the critically ill, only a small number of studies have specifically investigated its epidemiology in the intensive care unit (ICU). The objective of this study was to describe the occurrence of fever in the critically ill and assess its effect on ICU outcome. DESIGN Retrospective cohort. Fever was defined by temperature > or = 38.3 degrees C and high fever by > or = 39.5 degrees C. SETTING Calgary Health Region during 2000-2006. PATIENTS All adults (> or = 18 yrs) admitted to ICUs. INTERVENTIONS None. MEASUREMENTS AND MAIN RESULTS A total of 24,204 ICU admission episodes occurred among 20,466 patients; 35% were classified as medical, 33% as cardiac surgical, 16% as other surgical, and 15% as trauma/neurologic. The cumulative incidence of fever and high fever was 44% and 8% and the incidence density was 24.3 and 2.7 per 100 days of ICU admission, respectively. The incidence density of fever was higher in trauma/neuro patients, males, younger patients, and was lower in those with admission Acute Physiology and Chronic Health Evaluation II scores > or = 25. Seventeen percent and 31% of patients with fever and high fever had associated positive cultures. Resolution of fever and high fever occurred in 27% and 53% of patients before ICU discharge and prolonged fever and high fever lasting for 5 or more days in the ICU occurred in 18% and 11% of febrile patients, respectively. Although the presence of fever was not associated with increased ICU mortality (13% vs. 12%; p = .08), high fever was associated with significantly increased risk for death (20.3% vs. 12%, p < .0001). After controlling for confounding factors using multivariable logistic regression models, the influence of fever on the ICU mortality varied significantly according to its timing of onset, degree, and main admission category. CONCLUSIONS Fever is common in patients admitted to the ICU and its occurrence and impact on outcome varies among defined patient populations.