Metro trains often encounter the problem of wheel flange wear when they pass through sharp curves, especially for the poor wheel–rail profile matching or lack of lubrication, evidently increasing the maintenance and replacement costs for wheelsets and rails. Based on the field investigation and numerical simulation, the cause of severe flange wear of the original S1002 wheel profile on a Chinese metro line is analysed. An optimised wheel profile is proposed according to the rolling radius difference (RRD) for improving vehicle curving performance and mitigating wheel flange wear. A wheel wear prediction model is developed by combining the metro vehicle dynamics model considering multipoint contact with the long-term wear simulation model, and then validated with the field measurement results. The performance of the optimised wheel profile is evaluated numerically. The results indicate that the root cause of the wheel flange wear is the improper matching of wheel and rail profiles. The optimised wheel profile generally proves to perform better in terms of wheel–rail contact geometry, mechanical properties, vehicle running stability, curving negotiation and wear performance. The wheel flange thickness loss of the optimised wheel profile is decreased by 55.7% compared with that of S1002 profile.