All patients admitted to an Intensive Care Unit were randomised to receive all volume replacement fluid as either human albumin solution or a synthetic colloid. A total of 475 patients were admitted during the study period. Patients' age, sex, APACHE score and calculated risk of death were assessed on admission. Outcome was assessed as length of Intensive Care stay and mortality. There was no difference between the groups. Subgroups of patients with APACHE score greater than 10, calculated risk of death greater than 50% and length of stay greater than 5 days were also evaluated but not significant differences were found between treatment groups. The use of albumin rather than 3.5% polygeline for volume replacement in the Intensive Care Unit has no influence on outcome.