With emerging data that endothelial cell (EC) injury is the limiting factor in liver preservation and hepatic function, a simple and reliable biochemical technique for monitoring EC injury is needed. Measurement of purine nucleoside phosphorylase (PNP) release into the circulation from perfused liver has been proposed as such a method. However, our experiments with perfused rat liver did not display a clear or direct relationship between PNP release and endothelial cell injury. Therefore, we re-examined the suitability of using PNP as a measure of nonparenchymal injury by measuring its distribution in purified populations of hepatocytes, ECs, and Kupffer cells (KCs) and correlating cell injury and enzyme release in short-term cultures at 37±C of each cell type. Purified cells were incubated (4 × 106 cells/mL) in oxygen or nitrogen saturated Wisconsin solution or Krebs buffer for 6 hours, with cell viability and PNP release assayed every 2 hours. ECs had the lowest specific activity (27 ± 9 U/mg protein; mean ± standard error of the mean [SEM]) compared with both hepatocytes (115 ± 15) and KCs (66 ± 18). Despite a decrement in EC and KC viability over time in each incubation solution, there was poor correlation between time of incubation and PNP release (r = 0.01 to 0.22), and between cell viability and PNP release (r = 0.01 to 0.16). In contrast, PNP release from incubated hepatocytes correlated with the length of incubation (r = 0.57 to 0.78) as well as cell injury (r = 0.63 to 0.77) in all four test solutions. This data suggest that PNP release is unlikely to specifically reflect EC injury in the intact liver. (Hepatology 1995;21:174–179).