Statistical biases associated with single-level analyses underscore the importance of partitioning variance/covariance matrices into individual and group levels. From a multilevel perspective based on data from 21,579 students in 58 high schools, the present study assesses the multilevel factor structure of motivation and engagement with a particular focus on a recently developed instrument (the Motivation and Engagement Scale-High School; MES-HS) shown in a wide variety of contexts to reflect sound student-level psychometric properties. Multilevel (or hierarchical linear) confirmatory factor analysis using Mplus demonstrated that at student and school levels, the hypothesized framework comprised sound factor structure. Multilevel invariance tests further demonstrated that key parameters were invariant across student and school levels. Results also showed that first and higher order factor correlations were in hypothesized directions and broadly parallel at student and school levels. Taken together, the MES-HS target factors and the framework from which they emanate demonstrate sound multilevel psychometric properties. (Contains 2 tables and 1 figure.)