Teacher self-efficacy (TSE) is assumed to affect classroom environment, although empirical evidence for this assumption is still elusive. This study examines the relationship between TSE in behaviour management and classroom behavioural climate using a two-level cross-lagged model. TSE in behaviour management was assessed with six-item subscale of the TSE for Inclusive Practices scale while classroom behavioural climate was assessed with a 17-item scale measuring overall behavioural climate in the classroom. The longitudinal questionnaire data comprises responses from 1110 students and 75 class teachers who had not taught the target classes when the pre-assessment was conducted. At the time of post-assessment, the teachers had taught the target classes for a few months. This design may detect associations between TSE and classroom behavioural climate more sensitively than cross-lagged models were the main variables correlate already at the pre-assessment. Based on the analysis TSE in behaviour management significantly affected classroom behavioural climate, but not vice versa, when controlling the effect of several teacher characteristics and class-composition variables. This indicates that a connection between TSE and classroom behavioural climate seems to indeed exist.