Based on data from 153 early postmenopausal women who completed a double-blind, randomized 3 year study of graded hormone replacement therapy (HRT) doses or placebo, we investigated the value of bone markers to predict prevention of bone loss. Absolute values of serum and urinary CrossLaps (S-CTX and U-CTX) after 2 weeks of treatment were significantly correlated to 3 year bone mass response (r = -0. 28/-0.35; p < 0.001). These associations were fully expressed at 6 months (r = -0.61/-0.64; p < 0.001). Receiver operating characteristic analyses revealed that the predictive capacity of one measurement of a resorption marker after 6 months' treatment performed similarly as assessment of hip bone mass over 3 years in predicting preservation of spinal bone mass over 3 years. Comparable results were obtained using percent change from baseline in resorption markers at both 6 and 12 months, whereas for formation markers percent change was superior to absolute value at 6 months but not at 12 months. Values of accuracy for S-CTX for a cutoff of 1881 pmol/L at 6 months were 85.2% (sensitivity), 74.3% (specificity), 90.5% (positive predictive value), and 63.4% (negative predictive value); U-CTX performed similarly, whereas the values for the formation markers were slightly lower. A cutoff for S-CTX of 1245 pmol/L eliminated false-positive individuals (those who had a decrease below the cutoff but lost bone). In the false-negative group, which was composed of individuals whose S-CTX did not decrease below the cutoff but had preserved bone mass, S-CTX was significantly associated with spinal bone mass response (r = -0. 41; p < 0.01), indicating these women had been treated with a dose that was not at its optimum for their individual bone turnover. For this cutoff, the values were 49.5% (sensitivity), 97.1% (specificity), 98% (positive predictive value), and 40% (negative predictive value). In conclusion, early bone marker measurements predict long-term preservation of bone mass during HRT. Resorption markers seem superior to formation markers, which reflects that the primary effect of HRT is on bone resorption. A strategy with two cutoff levels may optimize the use of bone markers to predict bone mass response. Whether resorption markers can be used to guide individualized treatment remains to be investigated.