In the last thirty years, student engagement has received much attention as an important contributor to students’ school success. One major limitation of the research on student engagement is that there is not a widely accepted theory regarding what constitutes it and how it relates to motivation. In the present study I examined relations of college students’ motivational beliefs and task values (as defined in Eccles and colleagues’ expectancy-value theory, EVT) to proposed dimensions of their engagement: behavioral, cognitive, social, agentic, and behavioral and emotional disaffection. In particular, I examined: (1) empirical overlap among certain dimensions of engagement and task value constructs; (2) which EVT constructs are associated with which dimensions of engagement; (3) how motivational beliefs, values, and engagement dimensions relate over time; and (4) whether engagement dimensions mediate the relationship between motivational beliefs, values, and math and science grades. Students (Ntime1 = 486, Ntime2 = 516) were recruited from a large public university and then completed surveys about their motivation and engagement in their introductory math or science course twice, at the beginning of the semester and again after mid-terms examinations. Findings indicated that although there were strong associations among certain engagement dimensions and task value constructs, structural equation model fit indices indicated that these should be treated as separate constructs. Regression analyses showed that in general, students’ competence beliefs and values were associated with behavioral, cognitive, and emotional engagement and behavioral and emotional disaffection dimensions. However, the relations between the motivational variables and social and agentic engagement were weak or non-significant. Cross-lagged panel analyses indicated that some relations among task values and engagement dimensions were reciprocal over time, but more often motivation predicted engagement rather than the reverse. Students’ behavioral and cognitive engagement were strong mediators of the relations between their task values and domain-specific grades in math and science. I conclude from these results that (at least for college-aged students) certain engagement constructs should be integrated more fully into the well-established expectancy-value model; however, future research is needed to ensure that these relations hold across different domains.