Vertebral osteoporosis accounts for over 500,000 spinal fractures annually, the majority of which occur in older women. Despite these statistics, data regarding the rate of spinal bone loss in this population are conflicting. Moreover, the site of skeletal evaluation may significantly alter classification of osteoporosis in this age group. To examine trabecular-rich spinal bone loss with a measurement less affected by age-related artifacts than the AP spine, we measured lateral lumbar spine bone density (BMD) using dual-energy X-ray absorptiometry in 120 healthy, ambulatory, community-dwelling women 65 years of age and older (mean 70 +/- 5 years, range 65-88). We also examined cortical-rich sites in the forearm and total body along with AP spine and femoral BMD to assess the impact of site specificity using the World Health Organization (WHO) classification of osteoporosis. Significant losses in BMD were observed at the lateral spine (-1.1%/year, P < 0.01), forearm (-0. 77%/year, P </= 0.01), total hip (-0.75%/year, P </= 0.01), femoral neck (-0.70%/year, P </= 0.05), and trochanter (-0.78%/year, P </= 0. 01), but not the AP spine. Using the WHO criteria, lateral spine BMD determinations classified 66% of women with osteoporosis in contrast to 29% using the AP projection. Osteoporosis was diagnosed in 55% of women using measurements of the femoral neck, 43% using the total radius, and 19% using the total body. We conclude that elderly women lose bone at trabecular- and cortical-rich sites (lateral spine and total radius, respectively) in addition to sustaining significant age-related bone loss at mixed cortical/trabecular sites such as the hip. Classification of osteoporosis in this age group more than doubles using lateral versus AP spinal projections, supporting the necessity of developing more uniform agreement on site-specific analyses.