To identify predictors of short-term and sustained ALT normalization after interferon treatment in adult patients with chronic hepatitis C, we performed a meta-analysis of individual patients' data, with construction and cross-validation of a prediction rule, in 361 patients from two randomized trials. In one trial, 116 subjects with transfusion-related chronic hepatitis C were treated with lymphoblastoid interferon (5 MU/m2 three times a week for 2 mo, then 3 MU/m2 three times a week for 4 or 10 mo). In the other study, 245 patients with community-acquired chronic hepatitis C received recombinant interferon-alpha 2b (10 MU three times a week for 2 mo, then 5 MU three times a week for 4 mo; then random allocation of subjects with normal aminotransferase levels to stop or continue interferon for a further 6 mo). Overall, 164 subjects (45%; 95% confidence interval, 40% to 50%) had short-term responses; 61 (18%; 95% confidence interval, 14% to 22%) maintained sustained responses. Sixty patients (17%; 95% confidence interval, 13% to 21%) withdrew from treatment because of side effects or subjective intolerance. Logistic regression analysis showed that short-term and sustained response were independently predicted by lobular structure on pretreatment liver biopsy (p < 0.0001) and by short disease duration, defined as the time elapsed since transfusion in posttransfusion cases or since the first observation of abnormal aminotransferase levels in cryptogenic disease (p < 0.01). Rules to predict short-term and sustained response to interferon were derived from these items, showing a discriminatory ability of 0.73 and 0.70.(ABSTRACT TRUNCATED AT 250 WORDS)