The purpose of our study was to evaluate the effectiveness of self-tests as a component of web-based self-instruction in predoctoral orthodontics and pediatric dentistry. To this end, the usage patterns of online teaching modules and self-tests by students enrolled in three courses at the University of North Carolina at Chapel Hill School of Dentistry were monitored and correlated to final exam grade and course average. We recorded the frequency of access to thirty relevant teaching modules and twenty-nine relevant self-tests for 157 second- and third-year D.D.S. students during the course of our data collection. There was a statistically significant positive correlation between frequency of accessing self-tests and course performance in one course that was totally based on self-instruction with seminars and multiple-choice examination (Level IV): Spearman correlation between frequency of self-test access and final exam grade, rho=0.23, p=0.044; correlation between frequency of self-test access and course average: rho=0.39, p=0.0004. In the other two courses we monitored, which included content beyond self-instruction with self-tests, the correlations were positive but not statistically significant. The students' use of online learning resources varied significantly from one course (Level I) to the next (Level II): Wilcoxon matched pairs signed-rank tests, S=-515.5, p=.0057 and S=1086, p<0.0001. The data from this study suggest that increased use of web-based self-tests may be correlated with more effective learning in predoctoral dental education by virtue of the testing effect and that dental students' usage of resources for learning changes significantly over the course of their education.