Introduction: Differentiating between the components of wrist hyper-resistance post stroke, i.e., pathological neuromuscular activation (“spasticity”) and non-neural biomechanical changes, is important for treatment decisions. This study aimed to assess the reliability and construct validity of an innovative measurement device that quantifies these neural and non-neural components by biomechanical modeling. Methods: Forty-six patients with chronic stroke and 30 healthy age-matched subjects were assessed with the NeuroFlexor, a motor-driven device that imposes isokinetic wrist extensions at two controlled velocities (5 and 236°/s). Test-retest reliability was evaluated using intraclass correlation coefficients (ICC) and smallest detectable changes (SDC), and construct validity by testing the difference between patients and healthy subjects and between subgroups of patients stratified by modified Ashworth scale (MAS), and the association with clinical scales. Results: Test-retest reliability was excellent for the neural (NC) and non-neural elastic (EC) components (ICC 0.93 and 0.95, respectively), and good for the viscous component (VC) (ICC 0.84), with SDCs of 10.3, 3.1, and 0.5 N, respectively. NC and EC were significantly higher in patients compared to healthy subjects (p < 0.001). Components gradually increased with MAS category. NC and EC were positively associated with the MAS (rs 0.60 and 0.52, respectively; p < 0.01), and NC with the Tardieu scale (rs 0.36, p < 0.05). NC and EC were negatively associated with the Fugl-Meyer Assessment of the upper extremity and action research arm test (rs ≤ −0.38, p < 0.05). Conclusions: The NeuroFlexor reliably quantifies neural and non-neural components of wrist hyper-resistance in chronic stroke, but is less suitable for clinical evaluation at individual level due to high SDC values. Although construct validity has been demonstrated, further investigation at component level is needed.