The purpose of this study is to investigate the emprirical effect of four item presenting modes (i.e.list selection, random selection, flashcard queuing and variable intervalperformance queuing) on practice. Practice items used in this study are obtained from apaper-and-pencil test which consists of four options. To explore the effects of item difficulty on practice, three item banks, which vary in the statictical characteristics of items,are created. To implement these modes, presenting procedure are developed. The following conclusions emerge from the data analysis:1. No significance is observed for the interaction between the presenting modes and difficulty of practice items.2. The variable interval performance queuing is the most efficient one among the presenting modes used.3. Comparing the practice effect of the three item banks with different diffculties, the item bank with low difficulty shows significantly better effect than those with the moderate and high difficulty. The concrete results of the study would provide CAI designers and evealuators goodreference.1. The list selection mode and random selection node could be used when the practice items are with the same difficulty or condition. But the rehersal practices should be offerd when random selection is designed.2. The queuing presenting mode could be used when the items are with varied difficulties. And if there are a small number of items, the flashcard queuing strategy could be chosen to spare the time. For the variable interval performance queuing strategy, it could be used under most situations, but the interval of items should be shortened to save time.