A variable and person-centered approach was applied to understand the development of cross domain self-concepts of ability, patterns of math and English self-concepts of ability throughout adolescence, and their associations with college major. An expectancy-value perspective was integrated with dimensional comparison theory to understand how math and English self-concept of ability relate to one another over time and within a person. Regression analysis identified a positive association of math self-concept throughout adolescence on math related majors and a negative association of English self-concept on math related majors. Cluster analysis classified students into six to seven patterns of varying math and English self-concept of ability. Stereotypical gender differences were observed in cluster membership, with women overrepresented in high English clusters and males over represented in high math clusters. Cluster membership was predictive of the math related college majors. Students who were higher in math self concept of ability relative to English were overrepresented in math intensive majors. Findings support the importance of considering intraindividual hierarchies and ipsative relationships when studying the development of self-concept of ability and academic choices.