Nineteen patients with chronic hepatitis C who were virological non-responders (seven responder/relapse and 12 no response) to an initial 24-week course of interferon-alpha 2b (IFN-alpha 2b) at a dose of 3 million units (MU) thrice weekly were retreated for an additional 48 weeks at the same dosing schedule and followed-up for another 24 weeks post-therapy. At the end of follow-up (week 72), six (32%) of the 19 patients were hepatitis C virus (HCV) RNA negative and were virological complete responders to retreatment. The viral genotypes in these six patients included two each with 1b and 3a, one with 2b, and another with 2a/2b; five of the six virological responder patients had cirrhosis. Significant predictors for successful retreatment included lower baseline HCV RNA concentrations prior to the first course of therapy, 2 log10 reductions in serum HCV RNA during the initial treatment and classification into the virological 'responder/relapse' category after the first course of IFN (P < 0.01 for all observations). When the above factors were used to construct a predictive model to determine response to retreatment, it was found that the absence of a 2 log10 drop in HCV RNA concentrations during the first course of IFN therapy was the most reliable indicator of non-response to retreatment (likelihood ratio = 10, P = 0.0014). In addition, the presence of HCV RNA at week 12 of retreatment was 100% predictive of virological non-response to the 48-week course of therapy. Our findings indicate that an additional 48-week course of IFN-alpha 2b therapy at 3 MU thrice weekly will achieve a virological complete response in 60% of patients who had a 2 log10 drop in HCV RNA during their first course of treatment, and measurement of week-12 HCV RNA during retreatment to identify non-responders is beneficial to patients as well as being cost-effective. Thus, a second course of IFN remains a viable option in a subgroup of non-responder patients, regardless of genotype or the presence of compensated cirrhosis.