Teaching requires much emotion work which takes its toll on teachers. Emotion work is usually studied from one of two perspectives, a job or an individual perspective. In this study, we assessed the relative importance of these two perspectives in predicting emotional exhaustion. More than 200 teachers completed a questionnaire comprising the DISQ (Demand-Induced Strain Compensation Questionnaire), the Dutch Questionnaire on Emotional Labour (D-QEL), and the UBOS (Utrechtse Burnout Schaal [Utrecht Burnout Scale]). In line with previous studies, our findings indicated that emotional exhaustion is positively associated with emotional job demands and surface acting. The relative importance of the two operationalisations of emotion work was assessed by comparing the results of two regression analyses. Whereas the model with job demands explained 18% of the variance, the model with emotional labour explained only 5%. In understanding what might contribute to emotional exhaustion in teachers, the emotional job demands might be much more important than the self-regulation perspective that is measured with emotional labour. (Contains 3 tables.)