Although multiple theoretical traditions consider motivation, engagement, and learning as dynamic, these processes are often measured in more-or-less static ways in research. The study of student engagement exemplifies this mismatch well: Situational engagement is defined as context-sensitive, multi-dimensional, and dependent on momenyary factors such as teaching practices and peer influences. Yet, situational engagement is often studied using single time-point surveys, which can prevent researchers from examining the variation in situational engagement within person and across contexts that is presumed to exist. Our purpose in this study is to determine the extent to which students’ situational engagement (across cognitive, behavioral, and affective) varies at the situational, individual student, and classroom levels. To do so, we use intensive longitudinal methods and three extant datasets comprising 12,244 reports from 1,173 youth who were students in science classrooms. Our findings, at which we arrived using multivariate, mixed effects models, show that the greatest source of variation for situational engagement was attributable to individual students. Situational and classroom-related sources of variation were smaller, but still substantial. There were differences across the three dimensions of situational engagement, such that affective situational engagement was associated with greater variation at the situational level, cognitive engagement was associated with greater individual student variability, while behavioral engagement was positioned between the affective and cognitive dimensions with respect to its situational and individual variation. We discuss implications for modeling situations in studies utilizing intensive longitudinal methods.