The pyridinium cross-links of collagen pyridinoline (Pyd) and deoxypyridinoline (Dpd) are released during bone resorption and are neither metabolized nor absorbed from the diet. The aim of this study was to validate their use in osteoporosis. We studied 19 women with osteoporosis and estimated the bone resorption rate from a combined calcium balance/kinetics technique without (R) and with partial (R(H)) and "complete" (Res) correction for long-term exchange. The strongest correlation was observed between the bone-specific marker (Dpd) and with complete correction for long-term exchange (Res) (r = 0.71, p < 0.001). The intercept was not different from zero, suggesting that bone was the major source for Dpd. The crude ratio of Dpd to Res in the 19 women was 54.5; but the regression coefficient relating Dpd as the dependent variable to Res was 31.8 (95% CI 15.6-48.0), which was higher, but not significantly, than the ratio between Dpd and calcium (16.4) in 10 bone samples (cortical and trabecular bone). The weakest correlations between a biochemical marker and a kinetic index were those between hydroxyproline (a nonspecific marker of bone resorption) and R or R(H). Treatment with hormone replacement therapy (HRT) or HRT and parathyroid hormone peptide 1-38 in seven women over 1 year resulted in similar percent changes in the biochemical markers and estimates of bone resorption. We conclude that the measurement of Dpd provides a reasonably accurate assessment of bone resorption in osteoporosis, and in the context of several repeat 24-h collections of urine offers measurement precision that is similar to that obtainable with methods depending on the use of radioisotopic tracers and the assessment of metabolic calcium balance.