To compare the relative sensitivity and specificity of bone turnover indexes for bone loss or gain in early postmenopausal women, we performed a multicenter trial in 236 menopausal women (mean age, 51 yr), who were randomized to hormone replacement therapy (HRT) or calcium supplementation (CS; 500 mg/day) for 1 yr. Two markers of bone formation, osteocalcin (OC) and bone alkaline phosphatase (BSAP), and two markers of bone resorption, urinary N-telopeptide (NTx) and urinary free deoxypyridinoline (fDpd), as well as spine and femoral neck bone mineral density (BMD) were measured at baseline and 3, 6, and 12 months after treatment. Women receiving HRT (n = 105) showed a significant increase in spine BMD (+2.5%; P < 0.0001) and hip BMD (+1.0%; P = 0.02) compared to women receiving CS, who showed a decline at both sites (-1.1%; P < 0.01). All four markers showed time-dependent decreases in women receiving HRT (P < 0.001) and no change in women receiving CS alone. When baseline indexes of turnover were stratified by quartile, there was a significantly greater increase in BMD among those with the highest NTx, OC, and BSAP levels compared to that in those with the lowest NTx, OC, and BSAP levels (P < 0.05). The highest quartile for percent change from baseline to 6 months in fDpd, BSAP, and NTx was also associated with the greatest change in spine BMD at 1 yr. Receiver operator characteristic curves for percent change from baseline to 6 months in an individual marker to 1 yr change in BMD during HRT revealed that the percent change in NTx provided the greatest discrimination between gain and loss of BMD. When subjects receiving HRT were compared by their positive or negative skeletal response at 1 yr and their baseline turnover marker, initial NTx values were significantly higher in those that gained bone than in those that lost bone (P = 0.0002). CS women in the highest quartile for NTx at baseline had significantly greater decreases in spine BMD than subjects with the lowest NTx values (P < 0.005), although this was not true for fDpd (P < 0.20). In conclusion, for early postmenopausal women there are differential responses of biochemical markers to HRT and CS. Baseline urinary NTx and serum OC were the most sensitive predictors of change in spine BMD after 1 yr of either HRT or CS. Similarly, the percent change in NTx and OC from baseline to 6 months best predicted bone gain or loss. We conclude that markers of bone formation and resorption can be used clinically to predict future BMD in early postmenopausal women.