Percutaneous needle biopsy has many advantages over open biopsy in the treatment of neoplasms. However, the accuracy of the needle biopsy in the diagnosis of musculoskeletal tumors has not been established. It is essential to evaluate the accuracy and limitations of the procedure in musculoskeletal tumors. The diagnoses from 66 needle biopsies (bone, 37; soft tissue, 29) performed on 64 consecutive patients using a jamshidi needle (bone tumors) or a Tru-cut needle (soft tissue tumors) were compared with the final diagnoses made by open biopsy and/or a definitive operation. Fifty-eight specimens (87.9%) were judged to be adequate for histological examination. It was technically difficult to obtain undamaged cores from very hard bony lesions or sclerotic cyst walls. A pathologist with experience in musculoskeletal tumors was able to differentiate malignant tumors from benign lesions in 98.3% of the cases (bone, 100%; soft tissue, 96.4%) and arrive at a specific diagnosis in 91.4% (bone, 100%; soft tissue, 82.1%) when adequate cores were obtained. It was troublesome to distinguish a well-differentiated liposarcoma from a benign lipoma, or inflammatory lesions from benign tumorous conditions. The overall accuracy for needle biopsy was 80.3% (bone, 81.1%; soft tissue, 79.3%). There was no morbidity related to the procedure. The results indicate that meedle biopsy is a safe and accurate technique for diagnosing musculoskeletal tumors.