Ursodeoxycholic acid (UDCA) has been proposed as beneficial therapy for patients with primary biliary cirrhosis (PBC). The effects of UDCA on metabolic bone disease, a major source of morbidity in patients with PBC, are essentially unknown. Preliminary information suggests that UDCA may improve biochemical indices of bone disease, although information about the effects of UDCA on bone density is lacking. In this study, we describe the effects of UDCA on lumbar spine bone mineral densities over a 3-year period during which patients were enrolled in a randomized, double-blind, therapeutic trial of UDCA for the treatment of PBC. Lumbar spine dual-photon densitometry was measured at entry and annually. Eighty-eight patients, 50 in the UDCA group and 38 in the placebo group, had serial measurements available for up to 3 years. There was no statistical difference between the two treatment groups at entry with respect to histological stage, total bilirubin, age, use of calcium supplement, vitamin D levels, or estrogen. After 3 years of treatment, there was no significant difference in the lumbar spine bone densitometry measurements between the UDCA-treated and placebo groups. We conclude that, after 3 years of treatment, UDCA is not associated with statistically significant differences in the rate of bone loss from the lumbar spine in patients when compared with placebo despite beneficial effects of treatment on the underlying liver disease. Further efforts to define effective treatments for the bone disease need to be pursued.