Portal venous blood for rinsing out the University of Wisconsin solution (UWs) has the advantages of being a physiological fluid, removing acidotic mesenteric venous blood and perhaps resulting in more stable haemodynamic parameters during reperfusion. A group of 209 consecutive adult OLTs carried out between July 1993 and February 1995 were studied prospectively. The UWs was flushed out with 500 ml portal blood in 95 OLTs (group 1) and with 1.0 L 0.5% dextrose at 37 degrees C in 114 OLTs (group 2). The median day 1 and peak day 1-5 AST levels were significantly elevated in the 5% dextrose group: median 755 (118-11090) vs. 546 (121-6150) IU/I (P = 0.007, Wilcoxon); and median 1095 (159-11090) vs. 744 (157-7870) IU/l (p = 0.008, Wilcoxon), respectively. A median of 5 (0-27) units of blood were transfused in group 1 compared to 4 (0-54) units in group 2 (n.s.). There was no difference in peak bilirubin, lowest day 1-5 PT levels, primary nonfunction, median ITU stay, total inpatient stay and 1-month graft survival between the two groups (89% vs. 88%). Pre-reperfusion blood flush may be associated with less hepatocellular damage, without significant additional blood usage.