Students' emotions in the classroom are highly dynamic and thus typically strongly vary from one moment to the next. Methodologies like experience sampling and daily diaries have been increasingly used to capture these momentary emotional states and its fluctuations. A recurring question is to what extent aggregated state ratings of emotions over a longer period of time are similar to self-reported traits of emotions. Thus, this study aims to investigate the extent of similarity between students' aggregated emotional states and self-reported traits over a two-week period in three consecutive school years (N[subscript T1] = 149, average age[subscript T1] = 15.64 years). Six discrete emotions (enjoyment, anger, pride, anxiety, shame, and boredom) were assessed in German, English, French, and mathematics classes. We investigated similarity in terms of convergence, mean-level differences, long-term stability, and incremental predictive validity of aggregated states and self-reported traits. Results indicated substantial convergence between aggregated states and self-reported traits, with both showing similar long-term stability. However, aggregated states did not demonstrate superior predictive validity compared to self-reported traits for academic outcomes, while momentary assessments offer insights into short-term emotional fluctuations, on a person-aggregated level aggregated states and self-reported traits behave rather similarly. This suggests that both can be used interchangeably to study students' trait-related research questions, like interindividual differences or long-term emotional processes in educational settings.