We reviewed a consecutive series of 145 total hip arthroplasties that were performed between 1969 and 1983 in 113 patients under 40 years (mean age 33 years). Preoperative diagnosis was secondary osteoarthrosis in 49 hips, rheumatoid arthritis or ankylosing spondylos in 40, avascular osteonecrosis in 38, sequelae of septic arthritis in 6 and others in 12.All operations were performed in a standard operating room. A lateral approach was always used. Cemented implants were Charnley prosthesis (20) before 1973 and Charnley-Kerboull prosthesis (125) from 1973 to 1983.There were 2 delayed unions of the trochanter, 3 deep infections, 1 acute and 2 chronics. Recurrent dislocations due to excessive wear of the acetabular component were observed in 3 hips.At the most recent examination in 1994, 90 hips were reviewed, 48 hips had been lost to follow-up between 6 months and 15 years after surgery, 5 patients (7 hips) had died. The clinical and radiological results of the 137 hips followed up at least 2 years were available for the present study with a median follow-up period of 12 years (range 2 to 24 years). Of these 137 hips, 96 had been followed-up more than 10 years (mean 17 years, range 10 to 24 years).At the latest follow-up, the mean functional score in the d'Aubigné scoring system was 16,5 (range 8 to 18). There were 9 definite and 4 probable loosenings of the socket and 6 definite loosenings of the femoral component (5 subsidences and 1 fracture of the stem). The mean amount of the linear socket wear was 1,62 mm for the 137 hips and 2,22 mm for the 96 hips followed up more than 10 years. Cystic granulomatous lysis was observed in the acetabular area (14 hips), in the calcar (64 hips) and in other zones of the femoral cortex (16 hips). Heterotopic ossification was present in 43 hips.Of the initial 145 hips, 19 had been revised: 3 for septic loosening, 1 for fracture of the stem, 6 for excessive wear of the socket with periacetabular osteolysis and 9 for aseptic loosening (8 acetabular and 1 femoral component). Of these 19 revisions, 11 were performed after 10 years.According to these results the overall probability of retention of the prosthesis was 93,7 at 10 years, 81,5 % at 15 years and 77% at 20 years. When excluding the 3 revisions that were performed for septic loosening, the survival rate was 95,4% at 10 years, 84% at 15 years and 80% at 20 years.These results confirm that the long term reliability of total hip arthroplasty in young patients is less good than in older patients. The main cause of failure was aseptic loosening of the socket which was significantly correlated in this study with an excessive amount of wear. Nevertheless these results performed with a classical friction combination metal on polyethylene are similar or better than other reports with use of new friction combination as alumina-alumina. Further reseach will probably allow us to increase the wear resistance of total hip arthroplasty especially in young patients.Currently, the choices adopted in this series remain a good compromise for us that allow to expect good clinical and radiological results in more than 80% of the cases over 15 years.