We examined the efficacy of decreasing high doses (beginning at 18 MU/day) of interferon-alpha 2a vs. that of daily low doses (3 MU) in the treatment of chronic hepatitis delta virus infection. Patients treated with 18 MU had a somewhat higher frequency of normalization of serum ALT levels than patients treated with low doses (31% and 12%, respectively, on an intention-to-treat basis). A decrease in the percentage of hepatitis D virus RNA positivity was observed in both groups at the end of treatment. Thus, whereas in baseline samples 10 (62%) of the patients in each group were positive for hepatitis D virus RNA in serum on slot-blot hybridization, these numbers decreased to 5 (31%) and 4 (25%) patients in groups 1 and 2, respectively, at the end of therapy. However, hepatitis D virus RNA, detected by means of nested polymerase chain reaction, remained in all but two (one in each group) patients who completed the treatment. Finally, during posttreatment follow-up, hepatitis D virus RNA levels returned to baseline values, and only one patient remained negative for this marker. The beneficial effect of interferon-alpha was only transient. Only two patients (one from each treatment group) had persistently normal serum ALT levels after 18 mo of follow-up. Finally, the presence of serum hepatitis D virus RNA at the end of therapy, detected with nested polymerase chain reaction, might be a good marker for the prediction of viral replication relapse.