This observational study examined whether specific characteristics of 1168 patients admitted to a general intensive therapy unit had changed over 4 years. The patients' age, diagnosis, severity of illness on admission (APACHE score), duration of management and outcome were recorded prospectively. The risk of hospital mortality was calculated using the APACHE score modified by the appropriate coefficient for presenting diagnosis. The numbers of admissions increased over the study period from 240 patients in the first year to 344 in the last year. The pattern of diagnoses varied significantly with an increasing proportion of patients being admitted for postoperative observation in later years. There was a significant reduction in duration of admission with 70% of patients being admitted for < or = 3 days in 1988-89 while a similar proportion were admitted for a week in 1985-86. Approximately 40% of patients in each year were admitted with a risk of hospital mortality of < or = 10%; such patients received only a short period of intensive care and had a low mortality in the intensive therapy unit. A group of patients with a low predicted (and actual) mortality rate was identified; such patients may be more appropriately managed on a high dependency unit.