Expectancy-value motivation profiles were identified in a sample of US ninth-grade students in 2009 (n = 19,259) using latent profile analysis. Of four distinct profiles, two were high, one typical, and one low in math and in science. In each area, the two high profiles were distinguished by (1) high self-efficacy with lower utility value and (2) high utility value with lower self-efficacy. High-ability was identified by a math score at least one standard deviation above the mean within the race/ethnicity group. Forty-one percent of high-ability students had high math motivation, while only 27% had high science motivation. Evidence of disidentification was observed. Some high-ability students had low motivation in math (15%) and science (28%). Implications for talent development and gifted education are discussed.