Dimensional comparison theory (DCT; Möller & Marsh, 2013) assumes that students compare their academic achievement intraindividually across domains to form domain-specific self-concepts. Upward dimensional comparisons are believed to lead to lower self-concepts in the worse-off domain, while downward dimensional comparisons should lead to higher self-concepts in the better-off domain. Furthermore, DCT assumes the net effect of upward and downward dimensional comparisons to be beneficial to the self. To test these assumptions, 3 experiments and 2 field studies were conducted investigating the relative effects of upward and downward dimensional comparisons as well as their net effect. In Studies 1 (N = 149), 2 (N = 150) and 3 (N = 300), participants were asked to infer self-concepts of fictitious students after receiving experimentally manipulated information about their achievements in 2 domains, whereas participants in Studies 4 (N = 2,268) and 5 (N = 20,662) assessed their own self-concepts in German and mathematics. In all studies, downward dimensional comparisons resulted in higher self-concepts, whereas upward dimensional comparisons led to lower self-concepts. The net effect of dimensional comparisons was always found to be not statistically different from zero. The findings therefore support the central prediction of DCT on the discreteness of the effects of upward and downward dimensional comparisons, yet do not support the assumed positivity of their net effect. Furthermore, results indicate the effect patterns to be rather universal as they were stable across different samples, domains, achievement situations, research designs, and types of assessment.