The relationship between low bone mass and risk of fracture is well documented. Although bone densitometry is the method of choice for detecting low bone mass, its use may be limited by the availability of equipment, cost, and reimbursement issues. Improved patient selection for bone densitometry might increase the cost-effectiveness of screening for osteoporosis, a goal we sought to achieve by developing and validating a questionnaire based solely on patient-derived data. Responses to the questionnaire were used to assign postmenopausal women to one of two groups: (1) those unlikely to have low bone mineral density (defined as 2 standard deviations or more below the mean bone mass at the femoral neck in young, healthy white women) and therefore probably not currently candidates for bone densitometry; and (2) those likely to have low bone mineral density and therefore probably candidates for bone densitometry. We asked community-dwelling perimenopausal and postmenopausal women attending one of 106 participating multispecialty centers (both academic and community based) to complete a self-administered questionnaire and undergo bone density measurement using dual x-ray absorptiometry. We used regression modeling to identify factors most predictive of low bone density at the femoral neck in the postmenopausal group. A simple additive scoring system was developed based on the regression model. Results were validated in a separate cohort of postmenopausal women. Data were collected from 1279 postmenopausal women in the development cohort. Using only six questions (age, weight, race, fracture history, rheumatoid arthritis history, and estrogen use), we achieved a target of 89% sensitivity and 50% specificity. The likelihood ratio was 1.78. Validation in a separate group of 207 postmenopausal women yielded 91% sensitivity and 40% specificity. Assuming population characteristics similar to those of our development cohort, use of our questionnaire could decrease the use of bone densitometry by approximately 30%. Sensitivity and specificity can be varied by changing the level for referral for densitometry to provide the most cost-effective use within a particular healthcare setting. Thus use of our questionnaire, an inexpensive prescreening tool, in conjunction with physician assessment can optimize the use of bone densitometry and may lead to substantial savings in many healthcare settings where large numbers of women require evaluation for low bone mass.