Abstract Evaluation of graft quality remains a major problem in liver transplantation. The aim of this retrospective analysis was to examine the impact of donor criteria on postoperative graft function. Between June 1986 and September 1993 324 liver transplantations were performed at our institution. Criteria for exclusion from analysis were postoperative thrombosis of graft vessels, retransplantation, death prior to the 5th postoperative day or missing donor criteria. For the eligible 255 transplantations the impact of the following donor criteria were examined: age (range 1–62 years, median 28 years), sizejbody weigt index, duration of intensive care, cause of death, circulatory condition, need for vasopressive support and liver function tests (bilirubin, GOT, GPT, GGT, LDH, ALP, prothrombin time (PT), creatinine, sodium). The following intraoperative factors were also assessed: type of protective solution, cold ischaemic time (CIT), anhepatic period and blood transfusions. Graft function during the first 5 postoperative days was categorized into four groups: (1) good function (GOT max < 1000 U/l, spontaneous PT > 50%, bile production > 100 ml/day); (2) fair function (GOT 1000–2500 U/l, clotting factor support < 2 days, bile < 100 ml/day); (3) poor function (GOT > 2500 U/l, clotting factor support > 2 days, bile < 20 ml/day); (4) primary non-function (retransplantation required within 7 days). A univariate analysis revealed duration of intensive care (P= 0.001), circulatory condition (P= 0.005), anhepatic period (P= 0.0004), blood transfusions(P= 0.03) and CIT (P= 0.039) as significant risk factors for postoperative graft function. Entering these factors in a multivariate regression model we identified creatinine (P= 0.007), duration of intensive care (P= 0.009) and the size/body weight index (P= 0.03) as donor-related factors of high significance. Analysis of the intraoperative data revealed the anhepatic period as the factor of highest significance (P= 0.0004) together with CIT (P= 0.02) and intraoperative blood transfusions (P= 0.008). A doubling of the number of days of intensive care resulted in a threefold increased risk of postoperative graft failure. Prolonged intensive care is a variable representing multiple risk factors. Accepting donors with a longer history of hypotension of who show signs such as elevated creatinine should be carefully considered. In patients with expected surgical difficulties resulting in an extended anhepatic period and a higher blood loss, transplantation of organs retrieved from donors with a long duration of intensive care and a long CIT should be avoided.