To determine the factors associated with appendicular bone mass in older women. Cross-sectional analysis of baseline data collected for a multicenter, prospective study of osteoporotic fractures. Four clinical centers in Baltimore, Maryland; Minneapolis, Minnesota; Portland, Oregon; and the Monongahela valley, Pennsylvania. A total of 9704 ambulatory, nonblack women, ages 65 years or older, recruited from population-based listings. Demographic and historical information and anthropometric measurements were obtained from a baseline questionnaire, interview, and examination. Single-photon absorptiometry scans were obtained at three sites: the distal radius, midradius, and calcaneus. Multivariate associations with bone mass were first examined in a randomly selected half of the cohort (training group) and were then tested on the other half of the cohort (validation group). In order of decreasing strength of association, estrogen use, non–insulin-dependent diabetes, thiazide use, increased weight, greater muscle strength, later age at menopause, and greater height were independently associated with higher bone mass. Gastric surgery, age, history of maternal fracture, smoking, and caffeine intake were associated with lower bone mass (all P < 0.05). For example, we found that 2 or more years of estrogen use was associated with a 7.2% increase in distal radius bone mass, whereas gastrectomy was associated with an 8.2% decrease in bone mass. The associations between bone mass and dietary calcium intake and rheumatoid arthritis were inconsistent. Alcohol use, physical activity, use of calcium supplements, pregnancy, breast-feeding, parental nationality, and hair color were among the many variables not associated with bone mass. Multivariate models accounted for 20% to 35% of the total variance of bone mass. A large number of factors influence the bone mass of elderly women; however, age, weight, muscle strength, and estrogen use are the most important factors. *For a list of the investigators in the Study of Osteoporotic Fractures Research Group, see the Appendix. For current author affiliations, see end of text.