Background: The Box and Block Test (BBT) has been widely used to assess gross upper extremity (UE) motor function. We designed a haptic-combined virtual reality (VR) system, named VBBT, to improve BBT with kinematic parameters for more specific assessments.Methods: According to the block-transfer task in BBT, we designed a VBBT system with a haptic device to provide a sense of the grasping force and block's gravity during task performance for UE function assessment. Besides the number of transferred blocks (N-TB), some kinematic parameters were also collected in VBBT for more specific assessment, including the number of zero-crossings of moving acceleration (NZC-ACC), the number of zero-crossings of derivative of releasing force (NZC-DRF), the ratio of path length and linear length (PLR) and the distance between barrier partition and drop position (DDP). 16 post-stroke patients and 113 healthy subjects were recruited to examine quantitative performances, concurrent validity, test-retest reliability and subjective preference between the BBT and VBBT.Results: Based on the measurements of 95% healthy subjects, the normative ranges of quantitative performances were established for BBT and VBBT. Deficiencies in patients’ UE function could be identified when their measurements fell outsides. A moderate correlation was found in the N-TBs between the VBBT and BBT (r = 0.42). The measurements in VBBT presented a stronger age-related correlation than that in BBT (R2 = 0.57 and R2 = 0.16). The N-TBs in both BBT and VBBT were strongly correlated to the Action Research Arm Test (ARAT) (|r| = 0.84 and 0.83), and the NZC-ACC and NZC-DRF in VBBT also showed significant correlations. (|r| = 0.76 and 0.79). The N-TB, NZC-ACC and NZC-DRF in VBBT showed a good test-retest reliability (ICC = 0.75, 0.78 and 0.80) while a moderate reliability was found in BBT (ICC = 0.62). For the patient preference, VBBT was given a higher score than the BBT (p < 0.05) for its enjoyment and completion effort performing.Conclusion: The VBBT improved a gross manual assessment in BBT, which can provide clinically validated, reliable and motivative assessment with kinematic parameters for specific UE motor functions of post-stroke patients.