OBJECTIVE To determine the usefulness, limitations and diagnostic accuracy of fine needle aspiration biopsy (FNAB) in soft tissue tumours. METHODS Cross-sectional analytical (comparative) study. METHODS Department of Histopathology, Armed Forces Institute of Pathology, Rawalpindi. METHODS A total of 78 soft tissue tumours were subjected to FNAB from May 2000 to April 2002. Adequate aspirate was obtained in 69 patients. The smears were stained with Haematoxylin and Eosin (HE) and May-Grunwald-Giemsa (MGG) stains. FNAB smears were put into three categories i.e. benign, borderline or malignant. A definite diagnosis was also given where possible. Tissue biopsies were received in 38 of these patients. Routine HE staining was done in each case along with special stains and immunohistochemistry where required. RESULTS Out of 69 tumours, 30 were categorised as benign, 9 as intermediate and 30 as malignant. Lipomas were the most common benign tumour while small round blue cell tumours constituting 7 cases (23%) were the commonest malignancy. FNAB-histological correlation showed all tumours reported as benign on FNAB to be confirmed as such. There was only one false positive diagnosis on FNAB. All cases placed in intermediate category were found to be malignant on histopathology. Considering intermediate cases alternatively as benign and malignant revealed sensitivity of 80.6% and 100%, specificity of 85.7% and 85.7% and accuracy of 81.6% and 97.4% respectively. CONCLUSIONS Majority of soft tissue tumours can be categorized on FNAB with high degree of accuracy. The tumours placed in intermediate category should be subjected to biopsy or excision as it is more likely that they turn out to be malignant.