The ideal priming fluid for cardiopulmonary bypass is not known. We designed a study to determine whether there are important differences in the clinical effects of hydroxyethyl starch versus albumin when used in priming fluid, and in the clinical effects of colloid versus crystalloid priming fluid. We prospectively randomized 83 adult patients undergoing coronary artery bypass or valve replacement. All patients were managed by standardized protocol, and they received one of three priming fluids for bypass: hydroxyethyl starch (HES), 26 patients; albumin (ALB), 28 patients, and lactated Ringer's solution (LRS), 29 patients. The groups were stratified by body weight and type of operation. We measured 41 variables relating to operative time factors, fluid balance, bleeding, and organ function (renal, cardiac, and pulmonary) at several time intervals. The LRS group had a significantly lower colloid osmotic pressure than the other two groups, and the HES group had a substantially higher blood viscosity. Although the prothrombin time was significantly lower in the LRS group (p less than 0.05), the differences were very small and not clinically important. The platelet count in the HES group was significantly lower than in the other two groups immediately after bypass, but it was not different by the time the patients left the operating room. There were no differences among the groups in chest tube drainage, blood bank usage, or fluid balance. Postoperatively, the pulmonary shunt fraction was significantly greater in the LRS group. Body weight increased more in the LRS than in the HES and ALB groups (p = 0.01). No adverse reaction to the prime solutions was noted. The differences between the HES and ALB groups--prothrombin time, platelet count, and blood viscosity--had no apparent clinical effects; thus, the two may be considered clinically equivalent. The greater somatic and pulmonary fluid accumulation in the LRS group suggests that colloid is preferable to crystalloid in priming fluid.