The clinical presentation of alcoholic hepatitis (AH) can be mimicked by other alcoholic liver diseases. The aim of this study was to identify clinical features that predict AH on liver biopsy. Biopsies from patients hospitalized for presumed severe AH were used to identify a derivation cohort (101 patients) and validation cohort (71 patients). Using histologic scores for hepatocyte ballooning, Mallory-Denk bodies, and lobular inflammation, 95 patient biopsies (55%) were classified as definite AH, 55 (32%) as possible AH, and 22 (13%) as no AH. Survival was similar among the groups, but mortality was significantly increased for patients with fatty change ≤50% on initial liver biopsy. An analysis limited to uninfected patients with definite AH or no AH in the derivation cohort identified a greater leukocyte count at admission and radiographic evidence of liver surface nodularity as independent predictors of definite AH on biopsy (P < 0.05). In the derivation cohort, the leukocyte count thresholds for ensuring 100% specificity for diagnosing definite AH were 10 × 109/L if the liver surface was nodular and 14 × 109/L if the liver surface was smooth, with a sensitivity of 76% and an area under the receiver operator characteristic curve of 0.88. In the validation cohort, these thresholds had a specificity of 86%, a sensitivity of 59%, and an area under the receiver operator characteristic curve of 0.72. Conclusion: The combination of an elevated leukocyte count and a nodular liver surface in the absence of active infection retrospectively identified patients with a high likelihood of histologic AH for whom liver biopsy may not be necessary. For patients with suspected severe AH who do not fulfill these criteria, liver biopsy is important to exclude other variants of alcoholic liver disease. (Hepatology Communications 2017;1:1070-1084).