It has been known that biomechanical studies can gain insight of the causes and consequences of high loads on upper extremity during wheelchair propulsion (WP). Although a few research groups developed the instrumented wheel systems (IWS) to measure 3-dimensional (3-D) pushrim forces during WP for conventional wheelchairs, these IWSs are primarily designed for conventional wheelchairs, which have some limitations on applying to racing wheelchairs. To date, only one study developed the IWS that can measure pushrim forces during RWP, however, this system can measure only 2-D forces. Therefore, the purpose of this study was to develop and validate the IWS for measuring 3-D pushrim forces during racing WP. The IWS was consisted of a 3-D force transducer (45E15A-U760, JR-3 Inc, CA), an interface plate, a wheel, a pushrim, and a disc slip ring. The advantages of this IWS design are: 1) can be used interchangeably with 1.1 cm axles and 2) can be adapted with four different pushrim sizes. Static validation and dynamic validations were performed by compared the measured forces to the reference forces. For dynamic validation, the reference forces were obtained from the inverse dynamic model. Linearity, precision, maximum force deviation (MFD), and percent error were analyzed. To demonstrate a practicability of the IWS, an experienced racing wheelchair user propelled a racing wheelchair equipped with the IWS mounted on the right wheelchair hub on a stationary wheelchair roller. Kinematic data were acquired from six 120-Hz optical cameras while the IWS was connected to the A/D data acquisition board. Both kinematic and kinetic data were synchronized via TTL. Static validation showed a high linearity (0.910 <slope <1.410 and r² >0.999). The MFDs were found to be 1.97 N and 2.35 N, for F_z and F_xy, respectively while the highest percent error was found to be less than 3%. Under dynamic validation, the IWS provided the well-matched measurement forces with the predicted values from the inverse dynamic method.  (0.969 <slope <1.077 and r² > 0.811), but the MFD of dynamic and percent error rate were higher than the static condition. For RWP, the peak tangential, radial, and axial forces during were found to be 166.17 N, 139.02, and 48.67 N, respectively. The results suggested that the IWS can be used to obtain 3-D pushrim forces during RWP with the acceptable accuracy. The information from this IWS can help coaches not only to understand but may also improve RWP techniques.