Our previously reported multicenter, blinded, randomized, controlled study of two doses of interferon beta-1b (IFNB) in 372 patients demonstrated a reduction in relapse frequency and severity and in MRI activity. We now report the results of the continuation of that study. The median time on study was 46.0 months for the placebo arm, 45.0 months for 1.6 million international units (MIU), and 48.0 months for 8 MIU. IFNB had a persistent beneficial effect on exacerbation rate and MRI burden of disease and was relatively free of long-term side effects. There was a one-third reduction in exacerbation rate in the 8-MIU treatment arm, compared with placebo, in each of 5 years. Serial annual MRI was done in all patients, and 217 of the patients had either a fourth- or fifth-year scan. There was no significant progression of lesion burden in the 8-MIU arm in each successive year compared with baseline (at 4 years, p = 0.917), whereas a highly significant increase in lesion area occurred in the placebo arm (p = 0.0001). Among the 154 noncompleters, there was no systematic bias recognized that favored either treatment arm for the outcome measures of exacerbation rate, disability, or MRI activity. Dropouts in the placebo group had higher exacerbation rates and accumulation of MRI lesion burden than did dropouts in the other treatment arms, which probably reduced the power of the study to demonstrate treatment effects on these measures in the later years of the trial.(ABSTRACT TRUNCATED AT 250 WORDS)