The present study focused on an examination of both "global" and "domain-specific self-esteems" in secondary mathematics learning. The extent to which self-esteem, in general, would account and explain educational success through "social relationships with teachers" and "peers", and "personal interest in learning tasks" is the main inquiry of this correlational investigation. Two hundred and eighty-three year 10 students (128 girls, 155 boys) were asked to respond to a number of Likert-scale inventories. Causal modeling procedures, aided by the statistical software M"Plus" 7.3, were used to analyze the data and to test the hypothesized model. A series of a posteriori analyses yielded a modified model for discussion, producing a number of key findings, namely: (a) the differential influences of both global (e.g., [right arrow]relationship with teachers) and domain-specific (e.g., [right arrow]relationship with peers) self-esteems, (b) the positive influences of social relation with teachers (e.g., [right arrow]exam result) and interest in learning tasks (e.g., [right arrow]end-of-school term grade), and (c) the mediating functioning of relationship with peers and interest in learning tasks. This evidence, overall, provides additional theoretical insights into the operational nature and trajectories of effective learning.