OBJECTIVE The purpose of our study was to assess the accuracy of CT-guided biopsy of musculoskeletal neoplasms with respect to technique, anatomic site, and histology. METHODS During a 3-year period (January 1992 to December 1994), 176 core needle biopsies and 45 fine-needle aspirations were performed under CT guidance on patients with musculoskeletal neoplasms. To assess the accuracy of these procedures, we compared the diagnosis at biopsy with the final diagnosis as determined at the time of definitive treatment of the lesion. All biopsy findings were categorized as a primary malignancy (excluding round cell lesions), round cell lesion, local recurrence, or metastatic carcinoma. In addition, each lesion was analyzed according to which biopsy technique was used, whether frozen tissue section or rapid cytologic evaluation was used, and at which anatomic site the mass was found. RESULTS The accuracy for needle biopsy was 93% and that for fine-needle aspiration was 80%. The complication rate for both techniques was less than 1%. Accuracy rates for the four categories of primary malignancy, round cell lesion, local recurrence, and metastatic carcinoma were 87%, 75%, 94%, and 100%, respectively. The mismatch rates were similar in soft-tissue lesions (5/52) and bone lesions (16/169). Diminished accuracy was associated with round cell lesions (20%) and lesions located in the spine or the perivertebral region (20%). Nondiagnostic and insufficient specimens were found in 18 (8%) of the 221 patients. CONCLUSION. CT-guided biopsy of musculoskeletal malignancies is a safe and effective procedure if performed by a team of clinicians, pathologists, and radiologists who possess subspecialty expertise.