► Examined whether the specificity-matching principle extends to implicit self-concept. ► Participants completed measures of self-esteem and domain-specific self-concept. ► Used these measures to predict a global outcome and domain-specific outcomes. ► Implicit self-concept predicted specific outcomes, but not global outcomes. ► Implicit self-concept accounted for unique variance in the specific outcomes. According to the specificity-matching principle (Swann, Chang-Schneider, & McClarty, 2007), specific aspects of self-concept should predict domain specific outcomes, rather than broader outcomes. The purpose of the current study was to determine whether this principle, which has thus far been examined using explicit measures of the self, extends to the implicit self-concept. We tested this idea in the domain of math achievement. We observed that explicit math self-concept was correlated with specific outcomes (measures of math achievement), whereas explicit self-esteem was correlated with a broad outcome (satisfaction with life). Thus, we replicated the specificity-matching principle using explicit measures of self-esteem and self-concept. Moreover, we found that implicit self-concept was correlated with domain-specific outcomes, but not a global outcome, as the specificity-matching principle would predict. Furthermore, regression analyses indicated that implicit self-concept accounted for unique variance in the domain-specific outcomes, for which the other measures of the self could not account. Taken together, we conclude that the specificity-matching principle does indeed extend to the implicit self-concept.