BACKGROUND: In determining the level of bone resection in Ewing sarcoma, the most suitable time at which to perform magnetic resonance imaging (MRI) remains controversial. Current guidelines recommend that surgical planning be based on MRI performed prior to neoadjuvant chemotherapy. The goal of this study was to determine whether pre-chemotherapy or post-chemotherapy MRI provides greater accuracy of tumor limits for planning bone excision in the management of Ewing sarcoma. METHODS: This was a single-center, retrospective study. MRI was performed using 3 sequences: T1-weighted, T1-weighted with contrast enhancement by gadolinium injection, and a fluid-sensitive sequence (STIR [short tau inversion recovery] or proton-density-weighted with fat saturation). The tumor extent as assessed on pre-chemotherapy and post-chemotherapy MRI was compared with histological measurement of the resected specimen. RESULTS: Twenty patients with Ewing sarcoma of a long bone were included. In 6 cases, the tumor was located on the femur, in 5, the tibia; in 5, the fibula; and in 4, the humerus. The median patient age at diagnosis was 9.7 years. We found greater accuracy of measurements from MRI scans acquired after chemotherapy than from those acquired before chemotherapy. For both pre-chemotherapy and post-chemotherapy MRI, the greatest accuracy was achieved with the nonenhanced T1 sequence. There was no benefit to gadolinium enhancement. The median difference between T1 MRI and histological measurements was 19.0 mm (interquartile range [IQR], 4.3 to 32.8 mm) before chemotherapy and 5.0 mm (IQR, 2.0 to 13.0 mm) after chemotherapy. Adding a minimum margin of 20 mm to the limit of the tumor on post-chemotherapy T1 MRI always led to safe histological margin. CONCLUSIONS: Post-chemotherapy MRI provided a more accurate assessment of the limits of Ewing sarcoma. Surgical planning can therefore be based on post-chemotherapy MRI. Surgical cuts can be, at minimum, 20 mm from the limits as seen on MRI.