Sera from 735 patients with systemic sclerosis were classified according to antinuclear antibody (ANA) pattern as follows: centromere (25%), homogeneous (26%), fine speckled (21%), fine speckled with nucleolar (14%), coarse speckled (7%), nucleolar only (3%) and cytoplasmic only (3%). Immunoprecipitations using 35S-labelled HeLa cell antigen extract were performed using sera from 374 of these patients to detect the systemic sclerosis-specific antibodies to RNA polymerases I and III. The sera were selected to represent each ANA group, but focused on those giving fine speckled nucleoplasmic staining (with or without nucleolar staining) where all 86 sera positive for these antibodies were concentrated. Immunoprecipitates from a further 93 sera from patients with ANA-positive autoimmune diseases other than systemic sclerosis did not precipitate RNA polymerases. In addition, all sera were tested for antibodies to the extractable nuclear antigens topoisomerase I, nRNP, Ro, La and PM-Scl. Sera positive for antibodies to these antigens gave clear correlations with ANA patterns but, of the examples tested, none contained antibodies precipitating RNA polymerase I or III. Thus, sera containing antibodies to RNA polymerases I and III were exclusive of both anticentromere and anti-topoisomerase I, and formed a major serological subgroup (11.7%). Clinically, 77% were patients with diffuse cutaneous disease reflected by higher skin scores and a significantly higher incidence of renal involvement (33%) than patients with antibodies to topoisomerase I (3%).