This study examined the effect of different item feedback treatments (feedback and non-feedback) and delayed intervals (nine weeks, five weeks, and one week), as well as their interaction, on test scores of college students. Subjects were 90 undergraduates from two classes of the Department of Social Work, Chaoyang University of Technology, who enrolled in the course of Statistics for Social Work. Three pretests were administered through the semester. After each pretest, the items were randomly divided into two groups. Each class was then given feedback with different groups of items separately. During final exam, all items used in three pretests were put together as posttest items. Posttest scores were analyzed using two-way ANOVA for dependent samples, simple main effect test, and post hoc comparisons. The results showed that item feedback had effect in increasing test scores, but there were differences between two classes in the effect of different delayed intervals and the interaction. Suggestions were proposed for teaching and further research based on the results.