Study time allocation is an index to explore learners’ allocation of their attention and efforts. It contains item selection and self-paced allocation. According to the agenda-based regulation framework, learners’ study decisions are influenced by two qualitatively different kinds of processes: agenda based and habitual processes. Habitual process involves the task environment activating a prepotent or overlearned response. Previous researches only confirmed the influence of habitual process on learners’ item selection, but there was no influence on self-paced allocation. Since self-paced allocation was a dynamic process, therefore, This study used an eye tracking technique to investigate the effects of item position and difficulty on self-paced allocation. Forty students participated in the experiment. 3(Item difficult: easy, moderately difficult, difficult）×2(Item order: easy on the left [EMD] vs. difficult on the left [DME]) mixed factorial design was used, with item difficult as an within-subject factor. Participants were randomly assigned to either the EMD group (n=19) or the DME group (n=19). Participants were instructed that they were free to study any item in the study phase, We used an eye tracking technique to record the whole process of their self-paced allocation. After 16 study trials, participants completed an Arithmetic task and then received an cued recall test. The results showed that: (1) In the aspect of the whole study time, participants in both groups allocated the least time on easy items, and it was significantly difference than those of moderately difficult and the most difficult ones. (2) In the early of learning progress, There was significantly difference in the self-paced allocation to different levels of item difficult in the EMD group. People first allocated the most time to easy items, then to moderately difficult ones, but finally retried to the most difficult ones. The results were reversed for the DME group. Both item difficult and its position had effects on self-paced allocation. These results indicated that there was a habitual process in the early phase of self-paced allocation, which was in accordance with the agenda-based regulation model