Combining self-controlled and imposed levels of task difficulty during training seems to be better than completely self-controlled conditions. This study aimed at determining whether self-control of task difficulty affects skill learning during the early and late practice stages. For this purpose, 48 novice of high school students with a mean age (17.5 ± 1.75) years were randomly divided into four groups including complete self-controlled, self-controlled at the early, self-controlled at the end and paired. In the acquisition phase, which includes 100 trails,the self-control group was informed that they could choose and practice any of the preset distances from the target (1/5, 3, 3/5, 4, 4/5, 5 m) before each trial during the acquisition phase. The task difficulty was either self-controlled or imposed to the participants in the two phases of the acquisition session. The accuracy score of basketball throwing were analyzed by mixed variance analysis with repeated measures in acquisition and retention and transfer tests. The results showed that self-control of task difficulty improves the compared to the paired group (P=0. 001). Secondly, the performance of self-control of task difficulty in the early practice group was better than self-control of task difficulty in the end practice group (P=0. 001). Third, accuracy in the self-control of task difficulty in the early practice group was significantly better than self-control difficulty group and the paired group (P=0. 001). In general, the benefits of self-control of task difficulty were justified using the challenge point framework, the information processing perspective and cognitive perspective.