To evaluate cost-effectiveness and response predictors of treatment with recombinant interferon alpha-2a in chronic non-A, non-B hepatitis, 263 consecutive patients were enrolled in a multicenter long-term study. A pre-planned analysis aimed at identifying predictors of early response was carried out when all patients had completed the initial 3 months of treatment with 6 MU thrice weekly. Sixty-three percent of the patients enrolled were classified as responders. At multivariate logistic regression analysis, baseline gamma-glutamyltranspeptidase levels and cirrhosis were the only independent variables significantly associated with response. The risk of no response after 3 months of treatment was 3.9 times higher (95% confidence interval, 1.6 to 7.2) in patients with high baseline levels of gamma-glutamyltranspeptidase as compared with patients showing low baseline levels, and it was 2.0 times higher (1.1 to 3.8) in patients with cirrhosis as compared with those without it. We expect that results from this and other studies on large patient populations may help to select those patients who are more likely to benefit from interferon administration.