Student engagement is widely considered to be a multidimensional construct consisting of behavioral, cognitive, and affective components. Recent research has examined student engagement in science learning contexts using holistic approaches that account for multidimensionality through the identification of engagement profiles. However, it is not yet clear whether identified science engagement profiles are consistent across different samples, different learning environments, or different modes of measurement. Using data from three extant datasets involving middle‐ and high‐school aged youth, we explored the consistency of students' situational engagement profiles across science learning environments (formal and informal) and modes of measurement (experience sampling and end‐of‐class reports). Results from latent profile analyses of students' behavioral, cognitive, and affective engagement identified four common profiles across the three datasets, though each dataset had its own unique profile solution (ranging from five to seven profiles). Consistent profiles across the three datasets included universally low, moderate, recreational, and full engagement. Three additional engagement profiles were identified (rational, purposeful, and busy engagement), though they emerged inconsistently across the samples. Findings speak to the applicability of conceptual frameworks of engagement to science learning contexts. Results are discussed considering environmental (formal vs. informal learning environments) and methodological considerations (experience sampling vs. end of class report).