Forty patients with juvenile rheumatoid arthritis (JRA) were examined for C3b receptor (CR1) levels on erythrocytes (by an enzyme-linked immunosorbent assay), levels of circulating immune complexes (IC) (by a polyethylene glycol precipitation-complement consumption method), C3d split products (by intermediate gel rocket immunoelectrophoresis), and HLA-A, B, and C (by microdroplet lymphocytotoxicity test). Lower CR1 levels were found predominantly among patients with JRA (mean 57%) compared with 40 age-matched controls (mean 68%) (P = 0.008). The CR1 levels differed when the JRA patients were grouped by mode of disease onset and by clinical state at the time of testing (levels in patients with pauciarticular disease were higher than those in patients with polyarticular disease, and levels in patients with polyarticular disease were higher than those in patients with systemic disease) or by sex (girls had higher levels than boys; P = 0.01). The levels of circulating IC and C3d were elevated in 25% and 40% of JRA patients, respectively, and were mutually correlated (P less than 0.05, tau = 0.20). A negative correlation was found between levels of C3d and the numbers of CR1 (P less than 0.01, tau = -0.28), but concentrations of circulating IC did not correlate with CR1 values. CR1 levels were the same in 6 HLA-B27 positive and 25 HLA-B27 negative patients. These findings do not represent conclusive evidence that the number of CR1 on erythrocytes serves as a predictor of disease severity or as an indicator of disease activity in patients with JRA.