Recent science education reform efforts demand a more complex understanding of the quality of specific learning activities and the multidimensional nature of student engagement in science. In a unique application of person-oriented analysis of experience sampling data, we identify six distinct momentary engagement profiles representing different combinations of the behavioral, cognitive, and affective dimensions of student engagement in high school science classrooms. We examine the impact of student choice within learning activities, in particular the impact of framing, or choosing the problem or topic, within laboratory learning activities and find students spend a majority of their classroom time in one of several ‘intermediate’ engagement profiles characterized by high engagement on one dimension, but low levels on others. Laboratory activities during which students are more likely to engage in the scientific and engineering practices called for in the Next Generation Science Standards provided especially polarized engagement experiences, producing full engagement, universally low engagement, and pleasurable engagement in which students are affectively engaged but are not engaged cognitively or behaviorally. Student choice is generally associated with more optimal engagement profiles, but the type of choice matters: Choices about how to frame the learning activity have the most positive effects relative to other types of choices (such as with whom to work with). Results are discussed in terms of the importance of considering both characteristics of the learning activity and students’ engagement to understanding students’ experiences during science classes. Future directions related to the relationship between students’ in-class experiences and longer-term outcomes, such as their intention to study science during college or pursue a science-related career, are discussed.