The present recommendation is that aprotinin should be started before cardiac surgery, but as bleeding is only a problem in a minority, most patients are treated unnecessarily. In a prospective, randomised, double-blind trial we have studied the use of aprotinin, given only to the minority of patients who bled significantly post-operatively and who had not received prophylactic aprotinin. Sixty patients, who bled in excess of 400 ml in the first 3 h post-operatively were randomised to receive either aprotinin (2 x 10(6) KIU loading dose followed by an infusion of 0.5 x 10(6) KIU/h for 4 h) or placebo, in addition to conventional treatment. The demographic characteristics and the surgical procedures performed were similar in the two groups. Haematological variables were measured (A) before and (B) at the end of the infusion. Three patients were re-explored for excessive bleeding in each group and one patient died in each group. The patients in the aprotinin group bled significantly less and had higher haemoglobin levels on discharge than the patients in the placebo group. The tissue plasminogen activator antigen decreased and the fibrinogen level increased in the aprotinin group. In addition, aprotinin increased the number of surface GPIb platelet receptors as estimated by flow cytometry (36% versus 5%, P < 0.01) and maintained the platelet von Willebrand Factor activity (vWF). There was no significant difference in D-dimers, fibrin(ogen) degradation products, plasma vWF activity and antigen, platelet vWF antigen, platelet aggregation (to collagen, arachidonic acid, platelet activating factor and ristocetin), platelet count or transfusion of blood products between the two groups.(ABSTRACT TRUNCATED AT 250 WORDS)