In order to define the prognostic impact of the age of critical patients as well as its association with the initial severity of illness (SAPS index) and therapeutic effort (TISS index), we studied 1.102 patients older than 14 years admitted consecutively to 3 general intensive care units. The oldest patients (more than 65 years) showed a higher mean SAPS (11.00 versus 8.58, p less than 0.0001) and mortality (odds ratio = 1.99, p less than 0.0001) than the younger ones. The association between age and mortality showed a "dose-response" pattern, even after controlling for the effect of initial severity (p less than 0.0001). The effect of age was stronger in the groups of patients with low SAPS (odds ratio = 2.94, p = 0.0007) or admitted for acute myocardial infarction (odds ratio = 3.28, p less than 0.0001). The oldest group showed a lower TISS/SAPS ratio and a shorter stay in the intensive care unit than younger patients, suggesting a low relative therapeutic effort in the latter. The differential therapeutic effort did not explain, however, the mortality excess in the oldest group, because the gradient of mortality increased after adjusting for TISS/SAPS ratio (adjusted odds ratio = 2.42, p less than 0.0001).