The accuracy of the STAT-CRIT hematocrit (hct) was compared to Coulter and centrifuge methods in this study of the interrelationship between non-red cell blood constituents and accuracy of conductivity-based hct measurements. In the first part of the study, blood samples from 31 patients undergoing elective cardiac procedures were analyzed at three times: before induction of anesthesia (Time 1), during the rewarming period of cardiopulmonary bypass (CPB) (Time 2), and after transfusion of all cell-saver blood available after termination of CPB (Time 3). Laboratory evaluation included hct using the Stat-Crit, Coulter, and centrifuge methods, and sodium (Na), potassium (K), chloride (Cl), white blood cell count, total protein (TP), and albumin. In the second part of the study, patients were randomized to receive either 5% albumin (n = 14) or isotonic crystalloid (n = 14) after termination of CPB to determine the effect of protein colloid replacement on conductivity-based hct measurements. Blood samples were obtained before and after fluid volume replacement for multivariate analysis. Correlation of Coulter hct (absolute) with microhematocrit by centrifuge at all times (n = 93) was 0.95 (R2) with a bias and precision of -0.26 +/- 1.7%. Blood variables having the most significant effect on the Coulter-Stat-Crit difference (bias) were protein, Cl, and Na. Single regression analysis indicated that a 1-g/dL decrease in TP resulted in an absolute decrease in the hct reading by 1 hct% units. A 10-mmol/L change in either Cl or Na concentration resulted in a change in Stat-Crit accuracy of 3.5% and 2.5%, respectively.(ABSTRACT TRUNCATED AT 250 WORDS)