We have evaluated three commercial assays for collagen cross-links, two urine assays and a recently developed serum assay, as markers of bone turnover in 30 postmenopausal women with osteoporosis during their first year of treatment with the anti-resorptive drug alendronate. Before treatment, urine free deoxypyridinoline crosslinks (Dpd), urine N-telopeptide crosslinks (NTx) and serum C-telopeptide (CTx) values were within postmenopausal reference ranges. After 3 months' treatment the decrease in NTx and CTx was greater than that of Dpd (-50%, P < 0.0001 and -48%, P < 0.0001, compared with -11%, NS), as it was after 6 months (-51%, P<0.0001, and -57%, P<0.0001, compared with -19%, P<0.01). The decrease in resorption markers after 6 months was significant in 23% (Dpd), 66% (NTx) and 66% (CTx) of individuals. Neither baseline resorption marker values nor the per cent change after 6 months' therapy correlated with bone mineral density change (BMD) at either lumbar spine or femoral neck after one year's therapy, except baseline Dpd and lumbar spine BMD (P < 0.01). We conclude that NTx and serum CTx were more sensitive markers of bone turnover suppression by alendronate, but only baseline Dpd was useful in the prediction of individual bone density response after one year.