We have compared the findings of MRI with those of hip arthroscopy in 22 hips in adults with pilinful hip diseases. From March l995 to March 1997. We performed arthroscopy of the hips in symptomatic 22 patients. Fifteen cases were men and the mean average age was 46 years old. Preoperatively, the patients was diagnoseil with careful physical examinations and MR arthrogram or MR1. Among them, acetahular lahrum tears were 13 cases. synovial lesions were 3 cases and avascular necrosis of femoral head were 6 casee. We have examed cartilage softening, indentation or defect of femoral head and acetabulum, acetabular lahrum, loose hodies and synovial lesion hy arthroscopy. Under the arthroscopic guide, We had resected the torn Jahrum in S cases, removed a loose hodies such as synovial chondromatosis in 1 case and excised a synovial pathology in 2 cases. The others had a open arthrotomy in 8 cases. The sensitivity of MRI was 89.5% and its accuracy was 77.3% compared with arthroscopic findings. Average follow-up lengths was 8.6 months and 17(77.3%) of 22 patients were improved clinically. The arthroscopy of the hip is less invasive and a useful investigation for painful hip diseases for labral tears and synovial lesions in adults cspecially when standard MRI fails to prove a exact hip joint pathology, but it may be useless in avascular necrosis of femoral head. Also it is technical demanding and necessary to long-term follow-up.