We conducted a prospective, randomized trial to study the efficacy and tolerance of long-term versus short-term treatment with recombinant interferon alfa-2a in patients with chronic hepatitis B. Ten patients were randomly assigned to a 6-month interferon regimen, and 10 patients were assigned to a 3-week interferon trial. Eleven patients (five assigned to long-term treatment and six to short-term treatment) did not complete interferon therapy: eight had either severe thrombocytopenia or neutropenia; one had pronounced fatigue in relationship to administration of interferon; one had spontaneous bacterial peritonitis and sepsis and died; and one had a massive fatal variceal hemorrhage during interferon therapy. Most of the serious hematologic complications occurred in patients with cirrhosis and hypersplenism. In one patient, seroconversion to hepatitis B virus DNA negativity occurred before the onset of treatment. Four of the five patients able to complete the 6-month interferon regimen and only one of four patients able to complete the 3-week trial had seroconversion to hepatitis B virus DNA negativity. Thus, we conclude that the therapeutic response was better among patients who were able to complete a 6-month interferon trial. In patients with cirrhosis and hypersplenism, development of either severe thrombocytopenia or leukopenia associated with interferon therapy precluded completion of treatment.