A total of 754 consecutive patients with relapsing-remitting multiple sclerosis were investigated for interferon-beta (IFNbeta) antibodies by protein-G affinity chromatography and antiviral neutralization bioassay during 24 months on 6 MIU (22 microg) of subcutaneous IFNbeta-1a once weekly (n = 143) or three times weekly (n = 160), 6 MIU (30 microg) of intramuscular IFNbeta-1a once weekly (n = 140), or 8 MIU every other day of IFNbeta-1b (n = 311). The proportion of binding antibodies was higher in those receiving IFNbeta-1b compared with 6 MIU of IFNbeta-1a three times weekly (97 vs 89% at 12 months), and fewer became positive if 6 MIU of IFNbeta-1a was administered once weekly (58 vs 89%). Fewer patients on intramuscular than subcutaneous IFNbeta-1a became positive (33 vs 58%). The binding and neutralizing capacities were higher in the IFNbeta-1b group than in the IFNbeta-1a groups; these differences, however, were not significant after 12 months. The number of positive patients varied considerably and depended on the amount of IFN added to the bioassay; adding 10 LU/ml or more masked antibody detection. Antibodies induced by either preparation neutralized both IFNbeta species but not IFNalpha. In conclusion, IFNbeta-induced antibodies are frequently found in multiple sclerosis patients, and IFNbeta-1b is more immunogenic than IFNbeta-1a. The immunogenicity of IFNbeta-1a increases with the frequency of administration and if it is given subcutaneously.