Because it is unclear whether age-related bone loss results from increased bone resorption, decreased bone formation or both, we measured the serum level of bone Gla-protein (BGP), a specific marker for bone turnover, in 174 women, ages 30 to 94 yr. Serum BGP increased linearly with aging (r = 0.44, P less than 0.001) from 4.4 +/- 0.4 (mean +/- SE) in the 4th decade to 8.9 +/- 0.9 ng/ml in the 10th decade. This increase correlated inversely (P less than 0.001) with concomitant decreases in bone mineral density at the lumbar spine, midradius, and distal radius. Using partial correlation coefficients, serum BGP still correlated positively with age (r = 0.31, P less than 0.001) after creatinine clearance was fixed but not with creatinine clearance (r = -0.04, NS) when age was fixed. Urinary hydroxyproline (r = 0.29, P less than 0.001), an index of bone resorption, and serum alkaline phosphatase (r = 0.31, P less than 0.001), an index of bone formation, also increased with age and these increases correlated with increases in serum BGP (r = 0.39, P less than 0.001 and r = 0.43, P less than 0.001, respectively). Serum immunoreactive parathyroid hormone concentrations (r = 0.39, P less than 0.001) and urinary cyclic AMP excretion (r = 0.38, P less than 0.001) increased, suggesting that PTH secretion increased with age; these increases correlated significantly with increases in serum BGP. A subgroup of 32 women who were found to have vertebral fractures, hip fractures, or both had significantly higher values for serum BGP than the remainder. These data suggest that overall bone turnover increases in women with aging and, especially considering the concomitant decrease in skeletal mass, do not support the view that age-related bone loss results primarily from decreased bone formation.