BACKGROUND Preoperative assessment of orthotopic liver transplantation candidates requires definition of both the anatomy and metabolic function of the native liver. Current evaluation techniques combine computed tomographic scanning, duplex ultrasonography with blood chemistry analysis, and physical stigmata of end-stage liver disease. Recently, magnetic resonance imaging (MRI) has emerged as an alternative method for delineation of hepatic and portal venous anatomy. In addition, MRI accurately measures hepatic volume and portal venous blood flow. METHODS To examine the role of MRI-derived indexes of hepatic hemodynamics in the preoperative assessment of liver function, 39 consecutive liver transplantation candidates were studied in a prospective manner. Liver function (aspartate aminotransferase), alanine aminotransferase, alkaline phosphatase, total bilirubin, and albumin levels), hematologic indexes (complete blood cell count, prothrombin time), and Child's classification were determined at the time of evaluation. Axial breath-held multiplanar spoiled-gradient echo MRI measured hepatic volume, whereas a cine phase-contrast sequence perpendicular to the portal vein measured flow. RESULTS Hepatic index, defined as hepatic mass corrected for body surface area, was found to correlate with prothrombin time (p < 0.04) and platelet count (p < 0.03) by multivariate regression analysis. Portal flow index (PFI), defined as portal flow corrected for hepatic mass), was associated with aspartate aminotransferase (p < 0.02), alanine aminotransferase (p < 0.04), and albumin (p < 0.03) by multivariate regression analysis. In addition, PFI was closely correlated with the patients' functional status as determined by Child's classification system. Increasing values of PFI were associated with declining hepatic functional reserve. Child's class A patients had a mean PFI that was two times less than that of Child's class B patients (0.26 +/- 0.04 versus 0.04 +/- 0.06 ml/min/gm; p < 0.02) and five times less than that of Child's class C patients (0.26 +/- 0.04 versus 1.05 +/- 0.14 ml/min/gm; p < 0.001). Similarly, the mean PFI associated with Child's class B was two times less than that of Child's class C (0.46 +/- 0.06 versus 1.05 +/- 0.14 ml/min/gm; p < 0.01). These data show that MRI-derived indexes of portal hemodynamics and hepatic mass (1) correlate well with biochemical indexes of hepatic dysfunction and (2) serve as anatomic and hemodynamic correlates to Child's functional classification. CONCLUSIONS We conclude that MRI may serve to noninvasively delineate preoperative hepatic vascular anatomy and metabolic dysfunction in candidates undergoing examination for liver transplantation.