OBJECTIVES: BMD in the hand, as evaluated by digital X-ray radiogrammetry (DXR), has been suggested to be a predictor for joint damage in RA. A predictor for long-term prognosis might also predict increased mortality in RA. The aim of the present study was to evaluate BMD in the hand as a predictor for all-cause mortality. METHODS: In 1978, 152 consecutive patients (78% women, mean disease duration: 14.2 years) were enrolled. X-rays of the hands at inclusion were available in 108 patients. Reasons for not evaluating DXR in 24 patients were placement of joint prostheses or severe malalignment. BMD was evaluated by DXR on the same digitized hand X-rays used for scoring radiographic joint damage. Measures of disease activity and damage were used to predict mortality by Cox regression models. RESULTS: From February 1978 through March 2008, 62 of the 82 patients died, corresponding to a standardized mortality ratio of 2.92 (95% CI 2.19, 3.65) for both sexes combined. In age- and sex-adjusted proportional hazards models, BMD [hazard ratio (HR) = 0.58/1 s.d.; 95% CI 0.37, 0.91], Steinbrocker functional class 3-4 (HR = 4.74/1 step; 95% CI 1.93, 11.64), the physician's global assessment (HR = 1.38/1 s.d.; 95% CI 1.03, 1.84) and ESR (HR = 1.92/1 s.d.; 95% CI 1.42, 2.58) were significant predictors of mortality, but RF, disease duration, Larsen index, Ritchie articular index and the patient's global assessment were not. CONCLUSION: Low DXR-BMD predicted overall mortality in age- and sex-adjusted analyses, which further supports it as a valid measurement of disease activity or damage and as having prognostic value.