A randomized, double-blind, 1-year pilot study of prednisolone treatment for primary biliary cirrhosis was undertaken. Nineteen patients received 30 mg prednisolone per day initially, with a maintenance dose of 10 mg per day. Seventeen patients received placebo. The groups were matched for age, menopausal status, hepatic histological stage and bilirubin. Treatment was well tolerated without dropouts. Two patients receiving prednisolone developed diabetes, one a duodenal ulcer and one depression. One patient receiving placebo died for liver failure after 3 months. Cholestatic symptoms (itch and fatigue) improved on prednisolone. There was significant (prednisolone vs. placebo) improvement in transaminase (p = 0.0214), alkaline phosphatase (p = 0.0032), procollagen III peptide (p = 0.0103), immunoglobulin G (p = 0.0012) and liver histology (p = 0.016); these changes were greatest among noncirrhotic patients. No patient developed skeletal symptoms. Fifty-seven per cent had abnormal triolein breath tests prior to treatment, and 65% had abnormally low calcium absorption tests. Calcium absorption increased significantly in the treated group vs. placebo at 2 weeks (p less than 0.02), but not at 1 year. Femoral photon absorptiometry fell in the prednisolone group after 1 year (-3.5% vs. placebo +0.5%, p less than 0.05), as did trabecular bone volume (-6% vs. -2.8%, p less than 0.005) and resorption surface (-11% vs. +2%, p less than 0.02) on serial bone biopsy. Prednisolone seems to exert a favorable hepatic effect in primary biliary cirrhosis but at the expense of increased bone loss to approximately twice the expected rate. Prednisolone treatment merits further assessment in primary biliary cirrhosis over a longer period, with attention to selection of patients most likely to benefit and continuing observation of bone mass to better establish the "cost/benefit" ratio.