We have compared the analytical performance and biological variability of three commercially available bone resorption assays: Pyrilinks-D, Osteomark, and CrossLaps, for the measurement of urinary free deoxypyridinoline (Dpd), cross-linked N-telopeptides of type I collagen (NTx), and linear C-telopeptides of type I collagen (CTx), respectively. The intraassay and interassay CVs for precision of the Dpd and NTx assays were < 10% for analyte concentrations greater than the second calibrator (i.e., 3 nmol/L Dpd or 30 nmol bone collagen equivalents/L NTx). The CTx assay demonstrated poor precision for analyte concentration lower than the third calibrator (i.e., 200 micrograms/L). The NTx assay exhibited nonlinear recovery for sample dilutions prepared in buffer; however, this nonlinear recovery could be corrected for sample dilutions made in urine at a low analyte concentration. Supplement recoveries of each of the three assays were within 100% +/- 10% on average. All three analytes showed stability through five freeze-thaw cycles. The mean day-to-day variations were 16% for Dpd, and 23% for both NTx and CTx. Similar diurnal rhythm was observed for all three assays on average, with the peak in the early morning and the nadir in the afternoon. Mean amplitude of the diurnal variation was 37% for Dpd and NTx, and 57% for CTx. Variations within the reference intervals for a healthy premenopausal population were 28% for Dpd, 57% for NTx, and 56% for CTx. Pyrilinks-D has demonstrated analytical precision and accuracy equal or superior to Osteomark and CrossLaps in all areas. Dpd exhibits the least biological variability day-to-day, within individuals across the diurnal cycle, and within a healthy premenopausal population.