We have assessed the efficacy of cardiopulmonary bypass (CPB) using normal colloid oncotic pressure (COP) in a randomized, controlled study of 20 patients undergoing elective coronary artery surgery using heparin-coated circuits. For CPB, we used either crystalloid priming 1650 ml (n =10) or colloid priming 1650ml (2.4% modified fluid gelatin, n = 10). While COP did not change during bypass in the colloid group, a decline was observed in the crystalloid group (P = 0.005). By the end of bypass, the decrease in COP compared with baseline (ΔCOP) was 8.5 (S.D. 1.1) mm Hg in the crystalloid group compared with 1.5 (2.1) mm Hg in the colloid group (P = 0.0001). ΔCOP correlated positively with fluid balance during bypass (r2 = 0.41, P = 0.002). Similar increments in complement factors C3b/c and C4b/c, tumour necrosis factor-α and neutrophil elastase, but not endotoxins, were found in both groups as indicators of a systemic inflammatory response. A clinical performance score composed of fluid balance, post-operative duration of intubation and the difference between rectal temperature and skin temperature was more favourable in patients treated with colloid priming (P = 0.03). Median postoperative hospital stay was 7 (range 5-16) days in the crystalloid group compared with 5 (4-8) days in the colloid group (P = 0.016). Regression analysis indicated that CPB time, fluid balance during operation and postoperative PO2/F??O2 ratio were independent factors that predicted postoperative hospital stay. From these preliminary results we conclude that in the absence of endotoxaemia, use of a normal COP during CPB with modified fluid gelatin in heparin-coated circuits resulted in an improved post-operative course an a reduction in hospital stay.