A theoretical model linking achievement and emotions is proposed. The model posits that individual achievement promotes positive achievement emotions and reduces negative achievement emotions. In contrast, group-level achievement is thought to reduce individuals' positive emotions and increase their negative emotions. The model was tested using one crosssectional and two longitudinal datasets on 5 th to 10 th grade students' achievement emotions in mathematics (Studies 1-3: Ns = 1,610, 1,759, and 4,353, respectively). Multi-level latent structural equation modeling confirmed that individual achievement had positive predictive effects on positive emotions (enjoyment, pride) and negative predictive effects on negative emotions (anger, anxiety, shame, and hopelessness), controlling for prior achievement, autoregressive effects, reciprocal effects, gender, and socioeconomic status. Class-level achievement had negative compositional effects on the positive emotions and positive compositional effects on the negative emotions. Additional analyses suggested that self-concept of ability is a possible mediator of these effects. Furthermore, there were positive compositional effects of class-level achievement on individual achievement in Study 2 but not in Study 3, indicating that negative compositional effects on emotion are not reliably counteracted by positive effects on performance. The results were robust across studies, age groups, synchronous versus longitudinal analysis, and latent-manifest versus doubly-latent modeling. These findings imply that individual success drives emotional well-being, whereas placing individuals in highachieving groups can undermine well-being. Thus, the findings challenge policy and practice decisions on achievement-contingent allocation of individuals to groups.