Background Kinematic analysis is commonly used to objectively measure upper extremity movement performance after stroke. However, the concurrent validity and predictive validity of arm-trunk kinematics during reaching within and beyond arm's length have not been studied. Objective The aim of this study was to estimate the concurrent validity of kinematic measures before and after treatment and the predictive validity for reaching within and beyond arm's length after stroke. Design This was a secondary analysis study. Methods Ninety-seven participants with stroke (mean age=55.9 years [SD=10.9]) received intensive treatment every weekday for 3 to 4 weeks. Kinematic reaching tasks and the Wolf Motor Function Test (WMFT) were used before and after treatment. The validity of the kinematic measures was estimated in relation to WMFT scores. Results Of the 8 kinematic variables that were measured, index movement time before treatment (R2=.227–.362) and trunk movement time and trunk displacement after treatment (R2=.095–.346) had the strongest association with the WMFT at both reaching distances. Trunk movement time and trunk displacement before treatment explained 6.9% to 14.9% of the variance in the WMFT after treatment. Kinematic variables explained 6.9% to 49.3% and 9.4% to 38.7% of the variance in the WMFT during a task within arm's length and beyond arm's length, respectively. Limitations The study has limited generalizability. Conclusions Different kinematic variables may partially reflect motor function before and after treatment to a limited degree. Although the predictive validity was modest, trunk movement may be considered a prognostic determinant of motor function after treatment. A reaching task within arm's length may be a more suitable measure of kinematic performance for describing motor function than a reaching task beyond arm's length.