Background According to retrospective osteosarcoma series, ABCB1/P‐glycoprotein (Pgp) overexpression predicts for poor outcomes. A prospective trial to assess a risk‐adapted treatment strategy using mifamurtide in Pgp+ patients was performed. Methods This was a phase 2, multicenter, uncontrolled trial including patients 40 years old or younger with nonmetastatic extremity high‐grade osteosarcoma stratified according to Pgp expression. All patients received high‐dose methotrexate, doxorubicin, and cisplatin (MAP) preoperatively. In Pgp+ patients, mifamurtide was added postoperatively and combined with MAP for a good histologic response (necrosis ≥ 90%; good responders GRs) or with high‐dose ifosfamide (HDIFO) at 3 g/m2/d on days 1 to 5 for a histologic response < 90% (poor responders PRs). Pgp– patients received MAP postoperatively. After an amendment, the cumulative dose of methotrexate was increased from 60 to 120 g/m2 (from 5 to 10 courses). The primary end point was event‐free survival (EFS). A postamendment analysis was performed. Results In all, 279 patients were recruited, and 194 were included in the postamendment analysis: 70 (36%) were Pgp–, and 124 (64%) were Pgp+. The median follow‐up was 51 months. For Pgp+ patients, 5‐year EFS after definitive surgery (null hypothesis, 40%) was 69.8% (90% confidence interval CI, 62.2%‐76.2%): 59.8% in PRs and 83.7% in GRs. For Pgp– patients, the 5‐year EFS rate was 66.4% (90% CI, 55.6%‐75.1%). Conclusions This study showed that adjuvant mifamurtide, combined with HDIFO for a poor response to induction chemotherapy, could improve EFS in Pgp+ patients. Overall, the outcomes compared favorably with previous series. Mifamurtide and HDIFO as salvage chemotherapy are worth further study. The expression of ABCB1/P‐glycoprotein (Pgp) at diagnosis has been used to stratify patients with high‐grade osteosarcoma. Adjuvant mifamurtide, combined with high‐dose ifosfamide for a poor response to induction chemotherapy, can improve event‐free survival in Pgp+ patients.