Previous studies have indicated the human ability to learn different movements simultaneously, suggesting that motor learning can be achieved from partial use of attentional resources. However, the extent to which learning dual motor tasks is affected by the distribution of attention between tasks during practice is unknown. The aim of this study was to evaluate the effect of distribution of different proportions of attention to each component of a dual motor task during practice on its learning and automaticity. Thirty-five right-handed university students participated of the study. They performed the following tasks: drawing of a star-shaped figure based on inverted visual feedback with their left hand and sequentially touching the fingers with their right hand. The dual task consisted of the simultaneous execution of the two tasks. To evaluate movement's automatization due to practice, we used a probe reaction time task consisting of an auditory stimulus and a vocal response. Participants were assigned to one of three groups: practice of the dual task with asymmetric distribution of attention, allocating major attentional resources to the star drawing task and minimal attention to the fingers touching task (Dasy); practice of the dual task with symmetric distribution of attention between both tasks (Dsym); and practice of the single star drawing task (Sing). Results showed that after practice, the probing reaction time was reduced in the performance of the single and dual tasks in all groups, indicating augmented automaticity after practice. In the fingers touching task, the Dsym and Dasy groups outperformed the Sing group, indicating that this task was learned with reduced attention. In the star drawing task, the three groups improved similarly their performance after practice. Evaluation of the dual task revealed that both dual task practice groups, independently of the attention distribution regime, performed better than the Sing group after practice. No differences were found between the Dsym and Dasy groups across tasks, suggesting that they learned to a similar extent both tasks regardless of the attention distribution regime. These results suggest that performance improvement from practice, and associated movements automatization, can be achieved with allocation of partial attentional resources during the task acquisition.