The hemodynamic response to motor activation was investigated by time-resolved NIRS in healthy subjects and patients with unilateral impairment in motor ability. Healthy subjects performed a simple and a complex finger movement task, patients a handgrip task. A General Linear Model approach (GLM) was applied during NIRS data processing. In general, compared to the integral (continuous wave signal), higher significance of activation was found for the variance signal that selectively represents changes in the deep compartment. A discussion of GLM results with respect to task complexity and difficulty is provided.