Motor skill learning involves improvement in feedforward control, the ability to execute a motor plan more reliably, and feedback control, the ability to adjust the motor plan on the fly. The dependence between these control components and the association between training conditions and their improvement have not been directly examined. This study characterizes the contribution of feedforward and feedback control components to motor skill learning using the Arc Pointing Task (APT), a drawing task that requires high motor acuity. In experiment 1, 3 groups of subjects were tested with online visual feedback before and after training with online visual feedback (OF group), with knowledge of performance feedback that was presented after movement completion (KP group), and with both online and KP feedback (KP+OF group). While the improvement of OF group was not different from the improvement of KP+OF group, comparison of the KP and KP+OF groups revealed an advantage to the KP group in the fast test speed, suggesting that training without online feedback leads to a greater improvement in feedforward control. In experiment 2, subject's improvement was examined using test probes for estimating feedback and feedforward control. Both KP+OF and KP groups showed improvement in feedforward and feedback conditions with a trend toward a greater improvement of the KP group. Our results suggest that online visual feedback suppresses improvement in feedforward control during motor skill learning.