New movement assessment and data analysis methods are developed to quantify human arm motion patterns during physical interaction with robotic devices for rehabilitation. These methods provide metrics for future use in diagnosis, assessment and rehabilitation of subjects with affected arm movements. Specifically, the current study uses existing pattern recognition methods to evaluate the effect of age on performance of a specific motion, reaching to a target by moving the end-effector of a robot (an X-Y table). Differences in the arm motion patterns of younger and older subjects are evaluated using two measures: the principal component analysis similarity factor (SPCA) to compare path shape and the number of Fourier modes representing 98% of the path ‘energy’ to compare the smoothness of movement, a particularly important variable for assessment of pathologic movement. Both measures are less sensitive to noise than others previously reported in the literature and preserve information that is often lost through other analysis techniques. Data from the SPCA analysis indicate that age is a significant factor affecting the shapes of target reaching paths, followed by reaching movement type (crossing body midline/not crossing) and reaching side (left/right); hand dominance and trial repetition are not significant factors. Data from the Fourier-based analysis likewise indicate that age is a significant factor affecting smoothness of movement, and movements become smoother with increasing trial number in both younger and older subjects, although more rapidly so in younger subjects. These results using the proposed data analysis methods confirm current practice that age-matched subjects should be used for comparison to quantify recovery of arm movement during rehabilitation. The results also highlight the advantages that these methods offer relative to other reported measures.