Nonsteroidal anti-inflammatory drugs (NSAIDs) are known to inhibit synthesis of prostaglandins and may help prevent bone loss, but no study has shown the differential association of type or dose of NSAID compound with bone mineral density (BMD). The purpose of this study was to determine the relation of NSAIDs by type and dose to BMD. Participants were 932 Caucasian, community-dwelling women aged 44-98 years from southern California. Data were collected from 1988 to 1991 through the use of standardized medical questionnaires. Medication use was validated by a nurse. BMD at the ultradistal and midshaft radii were measured using single-photon absorptiometry, and at the hip and lumbar spine using dual-energy X-ray absorptiometry. Women (mean age, 72 years) were classified into 818 nonusers and 114 regular daily users of NSAIDs, of which 84 used propionic acid NSAIDs and the remainder used acetic acid NSAIDs. Occasional NSAID users were excluded. Women who used propionic acid NSAIDs, but not acetic acid NSAIDs, had higher BMD at all five sites and significantly higher BMD at the midshaft radius and lumbar spine. These differences remained after controlling for known covariates of osteoporosis. When women with self-reported osteoarthritis were excluded from the model, significantly higher BMD in propionic acid NSAID users was also observed at the femoral neck and total hip. Those who concurrently used estrogen and propionic acid NSAIDs had the highest BMD at all sites, suggesting an additive effect. We conclude that regular daily use of propionic acid NSAIDs, with or without simultaneous use of estrogen, may be helpful in preventing bone loss in older women. However, further research is needed to confirm these results before any clinical practice guidelines can be recommended due to the increased risk of serious complications associated with NSAID use.