Individuals’ punishment goals depend on the perceived cause of the misbehavior. However, a corresponding attributional model of punishment goals has only been studied in legal domains—but was largely ignored in others, such as the educational domain, in which student misbehavior and its treatment is a main stressor for both teachers and students. Thus, we investigated teachers’ punishment goals in classroom settings depending on their attribution of student misbehavior. Specifically, we asked laypeople (Experiment 1, N = 233), pre-service teachers (Experiment 2, N = 119), and in-service teachers (Experiment 3, N = 141) to read several versions of a scenario describing a student destroying the belongings of another student. Using a 2 x 2 within-subjects design, we manipulated the stability (stable vs. unstable) and controllability (controllable vs. uncontrollable) of the cause of the misbehavior. Results show that the support of retribution (i.e., evening out the harm caused) as a punishment goal in classroom interventions is largely independent of the perceived cause of the misbehavior. By contrast, the support of special prevention (i.e., preventing future misbehavior by the offending student) and general prevention (i.e., preventing future misbehavior by other students) is primarily subject to the perceived controllability of the misbehavior. Overall, this shows that models of punishment behavior developed in other domains cannot simply be applied to teachers’ classroom intervention strategies. This OSF project provides all instructions, materials, data, and results of our experiments (including supplementary analyses). Furthermore, we provide all information (materials & results) about our preliminary study that was conducted to test the materials used in our research.