Using data from TIMSS 2003, this study investigated the “big-fish-little-pond” effects on academic self-concept, and whether the effect size varies with age. Taking data structure into consideration, a two-level hierarchical linear model analysis was conducted. Data were collected from students who completed the mathematics self-concept items on student questionnaire and standardized math achievement tests that developed specially for TIMSS 2003. There were 4601 4th graders and 5355 8th graders. As predicted, results indicated the effect of individual student achievement on math self-concept were substantial and positive, .439 for 4th graders and .720 for 8th graders for linear component, .062 for 4 graders and .139 for 8th graders for quadratic term. The effect of school-average achievement on mathe self-concept are less substantial and negative, -.215 for 4th graders and -.261 for 8th grade. These effects are statistically significant and provide a strong replication of the contrast effect. As expected, a developmental change in the achievement-self-concept relation was supported. The interaction effect of individual student achievement and school-average achievement on math self-concept was only significant for 8th graders, but not for 4th graders. The negative effect of school-average achievement were slightly larger for students with high achievement levels.