Antibodies to Jo-1, a saline extractable nuclear antigen, were found in 11/47 (23%) patients with myositis and in 0/35 controls with systemic lupus erythematosus or progressive systemic sclerosis (p less than or equal to 0.01). Within myositis subgroups, anti-Jo-1 occurred in 6/20 (30%) with polymyositis, 2/16 (13%) with dermatomyositis and 3/7 (43%) with overlap. There was an association between Jo-1 antibodies and HLA-DR3, irrespective of race or clinical subgroups (p less than or equal to 0.05). All anti-Jo-1 positive patients had either HLA-DR3, HLA-DRW6 or both (p less than or equal to 0.01). Thus, this antibody system might arise as a result of an aberrant immune response mediated at or near HLA-D in patients with myositis.