To quantitate the inflammatory cell response within the portal tracts of liver allografts during acute rejection, we retrospectively and in a blinded fashion reviewed 431 biweekly, protocol, core allograft biopsies in 58 consecutive adult recipients. Following the determination of the cross-sectional area of each portal tract, the number of eosinophils, neutrophils, and lymphocytes therein was tabulated. The average number, percentage, and density of each type of portal inflammatory cell were calculated for each biopsy. Each biopsy was prospectively and independently classified as either associated (REJ+) or not associated (REJ-) with acute rejection. Acute rejection consisted of simultaneous allograft dysfunction and qualitative pathologic findings of acute rejection. Biopsies obtained during periods of normal allograft function or during episodes of dysfunction due to other causes were classified as not associated with rejection (REJ-). Ninety biopsies were classified as associated with acute rejection (REJ+) while 241 biopsies were classified as not associated with acute rejection (REJ-). In general, the average portal-tract number, percentage, and density of all inflammatory cells were significantly increased in biopsies associated with acute rejection. In contrast, only the portal-tract eosinophil values were consistently predictive of acute rejection following receiver-operating characteristic curve analysis (sensitivity = 82-86%, specificity = 91-92%). This quantitative method of allograft assessment appears to improve the objectivity of the serial biopsy protocol. By using this method, we found the eosinophil's appearance within the portal tracts to be a dependable indicator of acute rejection.