Open biopsy is recommended for a soft-tissue sarcoma (s-t-S) diagnosis. Core needle biopsy (CNB) was recently associated with minimal morbidity, cost and time-consumption, but also potential inaccuracy. Its diagnostic utility was investigated retrospectively in 110 patients with soft-tissue masses (s-t-M) undergoing CNB between September 1994 and September 2000. Sensitivity (Se), specificity (Sp), positive (PPV) and negative (NPV) predictive values were determined for malignancy (benign/malign), soft-tissue tumour (yes/no), and sarcoma diagnosis (yes/no), comparing CNB and the best standard test available; concordance was evaluated. 103/110 CNB were suitable for analysis. Final diagnosis was 23 benign tumours (19%), 65 s-t-S (59%), 9 lymphomas (8%), 6 fibromatoses (desmoid) (5%) and 7 carcinomas (6%). CNB Sp and PPV were 100%, Se was 95, 99 and 92%, and NPV 85, 95 and 88% for diagnosing malignancy, soft-tissue tumour and sarcoma. CNB Se and NPV were 100% for malignancy, connective tumour and sarcoma in lymphomas, high-grade sarcomas and desmoid tumours. In low grade sarcomas, Se was 94 and 85%, and NPV 84 and 77% for malignancy and sarcoma. Histological grade on CNB seems uneasy, except for grade-III tumours. CNB is accurate, not misleading for s-t-M diagnosis, avoids open biopsy complications, and allows one-surgery or neo-adjuvant chemotherapy planning when combined with appropriate imaging.