To examine the predictive value of biochemical markers of bone turnover for bone loss pre- and postmenopausally, we measured two markers of bone formation, bone-specific alkaline phosphatase (BALP) and intact osteocalcin (OC); four markers of bone resorption, urinary cross-linked N-telopeptides of type I collagen (NTx), type I collagen C-telopeptide breakdown products (CTx), hydroxylysylpyridinoline (HP) and lysylpyridinoline (LP); serum OC N-terminal (OC-N); and two serum cytokines, soluble interleukin-6 receptor (sIL-6R) and IL-1r antagonist at baseline and 1 year, as well as lumbar spine bone mineral density (BMD) at baseline and 1, 2, 3, 4, and 5 years after trial in 82 premenopausal (44.8 +/- 5.4 years old) and 325 postmenopausal (60.2 +/- 6.1 years old) healthy Japanese women. In premenopausal women, stratification of the baseline value of each biochemical marker into quartiles did not cause any significant difference in the change in BMD. Stratification of the NTx baseline value in postmenopausal women showed significant differences in rate of bone loss to the first year among those subjects with each quartile (Q1 [0.28 +/- 0.28%], Q2 [-0.32 +/- 0.34%], Q3 [-1.50 +/- 0.31%], and Q4 [-2.43 +/- 0.35%]) except for the difference between Q1 and Q2. The predictive value of NTx for BMD was greater in early postmenopausal women within 5 years after menopause than in late postmenopausal women with more than 5 years since menopause (YSM). Quartile analysis of the other biochemical markers and serum cytokines did not show any significant capacity for differentiating between bone loss rates. Moreover, when the changes in the lumbar spine BMD to the second and third years were stratified into quartiles by the baseline NTx, the ratios of bone loss to the second and the third years were significantly higher in those women with higher NTx (Q4; -3.15 +/- 0.56% and -4.06 +/- 0.57%, respectively) than in those with lower NTx (Q1; -0.74 +/- 0.44% and -1.03 +/- 0.51%, respectively). In conclusion, baseline urinary NTx was the most sensitive predictor of bone loss in the lumbar spine after 1, 2, and 3 years. Markers of bone resorption can be used clinically to predict future BMD in postmenopausal women.