Interferon-beta (IFN-beta) is currently the most commonly used treatment of relapsing-remitting multiple sclerosis (MS). At the time of this study, two preparations of IFN-beta were available, IFN-beta-1a (Avonextrade mark) and IFN-beta-1b (Betaferon(R)), which both can elicit an immune response with the development of anti-IFN-beta antibodies. Direct comparisons between these two preparations regarding antibody frequencies have, however, been difficult to perform, because two different analysis methods measuring partly different biological effects of IFN-beta have been employed. In the present study, binding and neutralizing anti-IFN-beta-1a and -1b antibodies were detected in parallel by an independent, well-acknowledged, interferon research laboratory using an immunoassay and a cytopathic virus inhibition assay. Five per cent of patients treated with IFN-beta-1a intramuscularly (n = 20) had neutralizing antibodies (NABs) compared with 44% of patients treated with IFN-beta-1b subcutaneously (n = 48). A high degree of cross-reactivity between neutralizing anti-IFN-beta-1a and -1b antibodies was observed. No effect of NABs on clinical outcome could be detected in this limited material. Binding anti-IFN-beta antibodies were observed in 20% of IFN-beta-1a treated patients compared with 81% of patients treated with IFN-beta-1b. Only one of 17 patients examined (6%) had detectable titres of binding anti-IFN-beta-1b antibodies in the cerebrospinal fluid (CSF). These data are the first using identical methodology to show that IFN-beta-1a gives rise to fewer NABs than IFN-beta-1b at recommended treatment schedules.