We report the results of a prospective Tunisian study using primary chemotherapy followed by conservative surgery in primitive limb osteosarcoma. From January 1988 to January 1998, 56 patients affected by limb osteosarcoma entered in a prospective study of neoadjuvant chemotherapy with the T10 protocol before surgery with a conservative intent. Initial work-up include: clinical exam with tumor measurements, chest and limb X-rays, limb CT-scan or MRI, chest CT-scan, bone scintigraphy and hematological and renal biological exams. Patients receive pre- and post-operative chemotherapy according to the T10 modified protocol. Fifty-six patients (33 M/23 F) with a mean age of 19 years (8 to 28) are included. Mean clinical and radiological tumor size is around 14 cm. Main histologic type is classic osteosarcoma (50% of cases) and 10 patients (9%) presented with initial metastasis; 42 patients on 56 receive the whole pre-operative protocol. Treatment is well tolerated excluding 18 episodes of mucositis, 29 of leucopenia (< grade 3), 7 of thrombopenia (< grade 3), 4 of cutaneous toxicity, 2 of pulmonary toxicity and 3 of nausea-vomiting. We observe 36% of good histological responders and 64% of bad responders to primary chemotherapy, 27 patients on 49 operated (53%) have a conservative surgery and 18 (47%) a radical surgery. With a median follow-up of 51 months (8 to 128), 29 patients remain alive free of disease (15/17 GR and 14/30 BR), 2 are alive with disease, 2 died by toxicity, 14 died by progressive disease and 9 are lost to follow-up with evolutive disease. Five year disease-free survival is 55% for the 46 non metastatic patients. In univariate analysis, seric alkaline phosphatase level (p = 0.0014) and histological response to chemotherapy (p = 0.0218) are significant factors for prognosis.