To determine the extent of low total-body bone mineral content (BMC) in non-corticosteroid-treated white postpubertal females with juvenile rheumatoid arthritis (JRA) compared with healthy age- and race-matched female controls, and to identify variables that significantly contribute to total-body BMC. Thirty-six females with definite JRA who had never received corticosteroids and 51 healthy female controls were evaluated. All subjects had had their first menstrual period at least 2 years prior to enrollment. Total-body BMC, lumbar spine bone mineral density, and body composition were determined by dual x-ray absorptiometry. Total-body BMC Z-scores were calculated for JRA patients using data from controls. JRA patients were dichotomized into those with "normal" bone mass (total-body BMC at or above the mean or no more than 1 SD below the mean) and those with "low" bone mass (total-body BMC more than 1 SD below the mean). Comparisons of anthropometric measurements, laboratory measurements of bone metabolism, disease activity, dietary intake, and physical activity were performed. Stepwise logistic regression was utilized to determine the presence or absence of low total-body BMC and to identify associated contributing factors. Total-body BMC was 4.5% lower in JRA patients than in controls (mean +/- SD 2,050 +/- 379 gm versus 2,143 +/- 308 gm; P = 0.21). Twenty-five of 36 patients (69.4%) had "normal" and 11 of 36 (30.6%) had "low" total-body BMC. Comparison of JRA patients with "normal" versus those with "low" total-body BMC revealed significant differences in disease characteristics, anthropometric and physical development characteristics, laboratory measures of bone mineralization, and dietary intake. The final regression model contained only lean mass (P = 0.01), which accounted for 76.3% of the variance in total-body BMC. The odds ratio for lean mass was 0.4451 (95% confidence interval 0.2374-0.8348). In this study, approximately 30% of the subjects in a sample of postpubertal female patients with mild-to-moderate, non-corticosteroid-treated JRA had low bone mass. The predictor variable that significantly contributed to total-body BMC was lean mass, which demonstrated a protective effect of 0.56 risk reduction for low total-body BMC.