Using digital X-ray radiogrammetry (DXR) on hand radiographs from a large population-based study, 1,370 postmenopausal women were evaluated in a prospective fashion; fracture occurrence was compared with DXR measurements of historic radiographs. Further, the aim of the study was to evaluate factors affecting DXR bone mineral density (BMD) in this cohort. The study is based on data from a subgroup of women participating in the third Copenhagen City Heart Study and additional data from a questionnaire obtained in 1999. The mean follow-up time was 6.1 years. During the observation period, 245 women suffered a fracture. Odds ratios (ORs) per 1 standard deviation decline in DXR-BMD were statistically significant for fracture in the groups of wrist fractures, proximal humerus fractures, vertebral fractures, and other fractures as well as in the total fracture group. In the hip fracture group, the P value almost reached significance (0.052). The highest ORs (2.4) were found in the group with proximal humerus fractures and in the vertebral fracture group (2.0). In the wrist fracture and hip fracture groups, ORs were 1.7 and 1.4, respectively. The group with other fractures had an OR of 1.7, and the OR in the entire fracture group was 1.6. Age, fracture, and smoking were negatively correlated with DXR-BMD, whereas BMI, age at menopause, hormone replacement therapy, and physical fitness and muscle strength were positively correlated with DXR-BMD. In conclusion, BMD estimated by DXR of the metacarpals predicts later osteoporotic fracture and seems to provide meaningful information on bone mass in epidemiological studies, where DXA measurements are not available.