This paper aims to identify risk factors for hip fracture in Medicare skilled nursing facility (SNF) residents and to develop a predictive model based on routinely collected administrative data (the Minimum Data Set, MDS) to identify high-risk residents. Prospective cohort study of 28,807 North Carolina Medicare SNF residents aged >65 years with a complete MDS assessment in 1999. Demographic, historical, physical, cognitive, behavioral, activities of daily living, and medication variables were obtained from the MDS. Hip fracture occurring after the first MDS assessment identified by ICD-9 code was the outcome measure. Variables significantly associated with hip fracture by chi-square test in a randomly selected derivation sample were combined in a multivariable logistic model and in models stratified by gender. The models were validated in the remaining subjects. Variables significantly related to subsequent hip fracture in the full cohort include: female sex (odds ratio 1.3, 95% confidence interval 1.0-1.7), white race (2.3, 1.6-3.5), age (1.03 per year, 1.01-1.04), cognitive impairment (1.4, 1.8-1.8), incontinence (0.68, 0.5-0.9), prior fractures (1.6, 1.2-2.1), and prior falls (1.4, 1.2-1.8). In ambulatory non-Hispanic white women, anxiety (1.5, 1.0-2.1), anxiolytic use (1.4, 1.1-1.9), wandering (1.4, 1.0-2.2), and training in community skills (1.4, 1.1-1.8) were new significant variables. For ambulatory non-Hispanic white men, education level (2.0, 1.2-3.2), weight loss (0.5, 0.2-1.0), history of osteoporosis (3.0, 1.3-6.7), pathologic bone fracture (9.7, 2.2-42.6), COPD (2.1, 1.3-3.5), glaucoma (2.6, 1.0-6.2), and standing balance impairment (1.8, 1.0-3.3) were also significant. All models were highly correlated with subsequent hip fracture, but the discriminative ability was limited (c statistic 0.678). Risk factors explained more of hip fracture risk in non-Hispanic white men (c statistic 0.793) than non-Hispanic white women (0.658). Risk factors for hip fracture in Medicare SNF residents have similarities and differences from those previously identified in community-dwelling older adults. Osteoporosis screening and intervention should focus on the healthiest, most independent subset of residents who have the greatest fracture risk.