Purpose: The purpose of this study was to investigate the effective learning method of non-dominant hand for the sequential timing performance. Methods: Twenty right-handed undergraduate and graduate students were randomly assigned to the non-dominant hand practice group that practiced with only non-dominant hand or the bimanual practice group that practiced with both hands simultaneously. The participant was asked to alternatively press two buttons six times with the index finger of the non-dominant hand or both hands in order to reproduce accurately the goal rhythm pattern (3,600ms in total duration). The goal rhythm pattern had three rhythm types and two perceptual structures, which the (in)congruent structure was that the structure of auditory signal was (or not) in accordance with that of visual signal. Results: Bimanual practice group was more effective for absolute-timing learning than non-dominant hand practice group. Although absolute timing performance improved in congruent structure during acquisition phase, the learning effect on absolute timing was occurred in incongruent structure. Relative timing performance of two groups improved in congruent structure during acquisition phase, but the learning effect on relative timing was occurred in both perceptual structures. Conclusion: We suggest that practice methods should be applied differently for absolute timing or relative timing, and the possibility of improving the learning effect of a non-dominant hand by utilizing an incongruent structure of visual-auditory information.