Chronic non-A, non-B hepatitis (NANBH) is a common and often progressive liver disease. Based on current serological tests, hepatitis C virus (HCV) infection is responsible for most cases. Interferon-alpha (IFN) treatment at a dose of 3 x 10(6) units given three times per week for 24 weeks has been shown to be effective in normalizing serum alanine aminotransferase (ALT) levels and reducing hepatic inflammation in approximately 40% of these patients. The purpose of this study was to identify pretreatment characteristics in patients with chronic hepatitis C(CH-C) which would best predict a favourable response to IFN treatment (normalization of serum ALT). One hundred and sixty-three adult patients who had participated in a large multicentre treatment trial were included in the study group; 84 had been treated with 3 x 10(6) units of recombinant IFN-alpha-2b (rIFN) subcutaneously three times per week for 24 weeks and 79 patients had been treated with 1 x 10(6) units rIFN in the same dosage schedule. Forty-one pretreatment historical, clinical, laboratory and histological variables were evaluated. In addition, early biochemical improvement during treatment was evaluated as a predictor of ultimate response. Univariate analysis identified six variables (dose, dose m-2, weight, body surface area, ongoing ethanol use, white blood cell count and the presence of symptoms) as potential predictors of response (two-tailed, P < 0.15). By multivariate analysis, however, only the 3 x 10(6) dose of rIFN was independently predictive of response (P < 0.01). When the analysis of response was confined to those patients who received treatment with 3 x 10(6) units of rIFN, seven variables [body weight, surface area, dose m-2, current ethanol use, serum albumin and the presence of chronic persistent hepatitis (CPH) on entry liver biopsy] were more frequent in patients who responded to therapy. In a multivariate model, only CPH and body weight predicted an increased likelihood of response (P < 0.01). However, the model was not a sensitive predictor of response as only 18% of the study group had CPH on liver biopsy. A decrease in serum ALT levels within the first 12-16 weeks of rIFN treatment was found to be the strongest indicator of an ultimate response to treatment. Thus, assessment of early response to IFN treatment is the only practical means of predicting complete response and avoiding prolonged and unnecessary therapy in those with little chance of response.