Two experiments were designed to examine the influence of the strength of perceptionaction coupling on the control and learning of a visuomotor tracking pattern. Participants produced rhythmic elbow flexion-extension motions to learn a visually defined 90° relative phase tracking pattern with an external sinusoidal signal which was set at 0.8 Hz with 8 cycles in a trial. Day 1 and Day 2 practice sessions consisted of a total of 72 practice trials. There were two visuomotor congruency groups, a congruent group with visual feedback representing the elbow’s rotation and an incongruent group with feedback representing the elbow’s rotation transformed by 180°. Before Day 1 practice (pre-practice) and 24 hours after Day 2 practice (post-practice), participants produced 0°, 45°, 90°, 135°, and 180° relative phase tracking patterns either with or without tracking feedback. The external signal and the limb’s feedback were provided in the same workspace in Experiment 1, while both signals were provided in a separate workspace in Experiment 2. The pre-practice results demonstrated that the 0° relative phase pattern was the most accurate and stable pattern, whereas the 90° and 135° relative phase patterns were less accurate and more variable. The incongruent group produced a more accurate and less variable 180° relative phase pattern compared to the congruent group. Practice led to a decrease in phase error and variability toward the required 90° relative phase pattern in both experiments. The congruent group produced more accurate tracking and less variable elbow amplitude compared to the incongruent group in the separate workspace, whereas no such congruency effects were found in the same workspace during practice. The post-practice results showed overall improvements in phase accuracy and stability in most relative phase patterns with practice. Overall deterioration in tracking performance was found when tracking without feedback in the pre- and post-practice sessions. These findings demonstrated that the perception-action coupling strength was modified by feedback, visuomotor mapping, perceptual pattern, and workspace framework. The differential strength of perception-action impacted the learning of the required visuomotor tracking pattern as well as the production of tracking accuracy and stability differentially among the other tracking patterns.