This study aims to understand post-stroke hand impairment using a hand exoskeleton robot and explore the possibility of using the device as a continuous monitoring tool to complement the currently used clinical scores. Six right-handed, right-side hemiplegic stroke survivors (55.8 ± 16.6 years old) and six age-matched neurologically-intact control subjects (55.3 ± 18.4 years old) were recruited. Each subject performed two repetitions of isometric flexion and extension maximum voluntary torques (MVTs) on each finger individually and all five fingers together in a randomised order. All joint moments were recorded. A finger individuation index (FI) was derived from the joint moments data. Correlation between the average of the FIs, namely FII, and the clinical scores was then evaluated. The results suggest that stroke subjects produce significantly lower joint moments, about one third of that of the control subjects. Significant differences in FIs were found on all fingers except the thumb, with those of the control subjects ranging from 0.5 to 0.7 compared to the 0.2 to 0.4 of the stroke subjects. FII values were found to correlate well with different clinical scores. Stroke survivors were found to be restricted by lack of both strength and individuality in their fingers. Additionally, FII could potentially serve as a continuous monitoring system of finger individuation. ; Interdisciplinary Division of Biomedical Engineering