Background: According to the classroom ecology paradigm, teachers and students interpret, predict, and respond to each other repeatedly in a reciprocal way. Such a reciprocal relationship is reflected in bidirectional interactions between a teacher's behavior and student (dis)engagement, an issue that has been confirmed in longitudinal studies including measures at different moments in a school year. Aims: Starting from the perspective of self-determination theory, the aim of the present study was to investigate bidirectional relationships between student (dis)engagement and need-supportive and need-thwarting teaching behavior during the first 15 min of a lesson. Sample & method: The first three 5-minute intervals of 100 videotaped physical education lessons taught by 100 different teachers (51.9% male, Mage = 37.5 + 10.9 years) were observed and coded for need-supportive and need-thwarting teaching behavior, student engagement, and student disengagement. Correlations were calculated to explore relationships between student (dis)engagement and teaching behavior over the first 15 minutes of a PE lesson. Next, path analyses were conducted to analyze 5-to-5 minute interactions between teaching behavior and student (dis)engagement. Results: Student engagement correlated positively and disengagement correlated negatively with need support, while engagement correlated negatively and disengagement correlated positively with need-thwarting over the first 15 minutes of the lesson. There were few significant relationships between student engagement and teachers' behavior across and between each of the three 5-minute intervals. Only when teachers provided more need support during the first 5 minutes of the lesson, students were more engaged in the third 5 minutes of the lesson. When students were more disengaged during the first 5 minutes of the lesson, teachers displayed less need support in the following 10 minutes of the lesson. In contrast, student disengagement in the second 5 minutes of the lesson related to more need support in the next 5 minutes. Most of the within-interval relationships between student engagement and teachers' behaviors were inconsistent, but we did find positive relationships between student disengagement and need-thwarting teaching behaviors in the first and third interval, suggesting a rather direct and momentary within 5-minute intervals interaction between teachers and students. Conclusions: Findings of the present observational study suggest that, although overall relationships between student (dis)engagement and teachers' behavior were in the expected directions, the picture might become more complicated when relationships are investigated according to the timing of the lesson, an issue that has remained uncovered in self-reported studies. While student disengagement was related to less need support and more need-thwarting teaching behaviors, more detailed analyses showed that it was particularly student disengagement in the beginning of a lesson that elicited less positive teaching behaviors. When students display disengagement further along in the first 15 minutes of the lesson, teachers seemed to respond in a more need-supportive way to student disengagement. Such findings provide interesting insights to build interventions for teachers around certain critical moments during the lesson, for example when dealing with student disengagement at a specific moment in the lesson.