Deflazacort (DFZ) is claimed to have fewer adverse bone effects than prednisone (PDN) at doses with equivalent anti-inflammatory activity (5 mg PDN = 6 mg DFZ). However, its safety over the long-term has never been tested in a controlled trial. The aim of the present study was to assess prospectively the safety and efficacy of DFZ compared with PDN in previously untreated patients with chronic, histologically proven sarcoidosis needing long-term (> or = 2 yr) corticosteroid therapy. Thirty-six patients were treated with PDN for 32 +/- 18 months and 36 patients were treated with DFZ for 42 +/- 18 months, and followed-up with periodic chest X-ray, 67Gallium lung scan, angiotensin converting enzyme (ACE), serum and urinary calcium levels, spirometry, alveolar diffusion (DLCO) arterial oxygen tension (PaO2), bone mineral content (BMC) (by computed tomography), and a complete biochemical and haematological profile. The two groups were similar as regards sex, age, pulmonary and extrapulmonary involvement, parameters of activity and impairment, and initial BMC. Daily starting doses were 23.2 +/- 11.4 mg DFZ and 22.3 +/- 6.9 mg PDN. One year of trial was completed by 69 patients, 2 yr by 59 patients, 3 yr by 46 patients and 4 yr by 24 patients. Some patients were followed-up for 5-7 yr. The mean daily dose over the whole period was 15 +/- 10 mg DFZ and 10 +/- 6 PDN, starting from 21 +/- 9 and 15 +/- 8 mg in the first year, and progressively declining to a mean of 9 +/- 6 mg in both groups in the fourth year. Chest X-ray, 67Ga score, ACE and forced vital capacity improved significantly in both groups. Urine total calcium improved significantly in the PDN group (345 +/- 27 to 186 +/- 47; P < 0.05) with a similar but non-significant pattern in the DFZ group (270 +/- 28 to 207 +/- 39). Non-significant improvements were observed in DLCO, PaO2 and forced expiratory volume in 1 s in both groups. Drug-related adverse events were more frequent in the PDN group, causing discontinuation of the drug in four PDN patients. Body weight increased mainly in the PDN group [69.9 +/- 0.4 to 73.6 +/- 0.8 kg vs 70.1 +/- 0.4 to 70.0 +/- 0.6 kg in the DFZ group (P < 0.01)]. Bone mineral content dropped under the fracture threshold in most PDN patients, who thus appeared at higher risk for fractures. In fact, six atraumatic skeletal fractures were observed in this group but only one in the DFZ group. Two further patients in the DFZ group and eight in the PDN group were obliged to start corrective measures for bone loss and/or bone pain. At the end of the study, 21 patients (12 DFZ, nine PDN) no longer needed corticosteroids, and the others were taking a maintenance daily dose that controlled the disease adequately. In conclusion, DFZ appeared as effective as PDN in the long-term treatment of chronic sarcoidosis, and it may have fewer side-effects, especially on bone.