The types of intensive care are multiple. The aim of this multicentric study was to describe activity of different ICUs using the same methods. 38 ICU were chosen by cooption, not randomization. Collected data concerned input (age, previous health status (HS), Simplified Acute Physiology Score or SAPS, Intensive Care Group (ICG), processes (TISS points), percentage of ventilated patients and pulmonary arterial lines and outcome (ICU death rate). The 3 ICG were: M = medical: all the none surgical patients; S = surgical patients operated in emergency setting during the week preceding or following ICU admission; E = surgical patients whose admission to ICU was scheduled at least 24 h before because of elective surgery. 3,687 patients were studied, classified as follows: M = 2175; S = 885; E = 627. The first part of the results concerned the differences between the three ICG: inputs, processes and outcome were very different in the three groups M, S, E, particularly in the E (elective) group, where therapeutic level was higher for low SAPS and mortality lower for high SAPS. The second part of the results concerns the differences between the ICUs. Intermediate units had older, less severe, and mainly medical patients. Surgical patients had better previous health status, were younger and scheduled for 40%. TISS points were higher, mainly by a higher rate of ventilated patients and patients with pulmonary artery lines on the first day. Specialized units characteristics depended mainly on the ICG.(ABSTRACT TRUNCATED AT 250 WORDS)