The present research is the first to integrate three major theories of academic self-concept formation into one framework: the big-fish-little-pond effect (BFLPE), the basking-in-reflected-glory effect (BIRGE), and the internal/external frame of reference (I/E) model. For separating the BFLPE and the BIRGE, we introduced students' collective self-concepts, defined as their perceptions of domain-specific abilities of their group (e.g., their class). We tested our newly developed model, predicting students' math and verbal self-concepts, in the field (N = 2,001 students from 110 classes) and in the laboratory (N = 315 students in groups of three). In both studies, we found support for (1) the BFLPE (i.e., negative effects of group-average achievements on individual self-concepts within domains), (2) the BIRGE (i.e., positive effects of collective self-concepts on individual self-concepts within domains), and (3) the I/E model (i.e., positive/negative effects of individual achievements on individual selfconcepts within/between domains). Moreover, for the first time, we have largely replicated the I/E model relations for predicting students' collective self-concepts by their group-average achievements. Interestingly, the BIRGE exceeded the BFLPE in both studies. This finding demonstrates the high suitability of collective selfconcepts for operationalizing the BIRGE. Overall, our findings substantially enhance our knowledge of the cognitive processes involved in students' self-concept formation.