Objective: To evaluate the impact of three types of social cognitive theory (SCT)-based elementary school classroom physical activity (PA) training on teachers' implementation rates, attitudes, knowledge and behaviour. Design: Key stakeholder focus groups informed development of phase II which took the form of a randomised controlled trial of three different intensities of teacher training to conduct classroom-based PA sessions. Setting: The study was conducted over one school year (2016-2017), across four elementary schools in the USA. Methods: Researchers delivered professional development to teachers, focusing on effective methods for PA use in the classroom through three formats: webinar, in-person training and in-person with personalised assistance. Training content was built on SCT constructs. This study examined the impact of the PA training on SCT construct outcomes among elementary school classroom teachers (n = 41). Study outcome measures included reciprocal determinism, behavioural capability, outcome expectations and self-efficacy. Over the 12-week period following the training, teachers completed surveys to measure SCT construct outcomes. Results: Results of the intervention demonstrated an increase in teacher self-efficacy and knowledge (p < 0.01 and p < 0.01). There was a significant increase of teachers who started using PA in the classroom who indicated no use at baseline (from 21% to 6% not using). Conclusion: The use of SCT to provide a teacher training is an effective method to increase classroom PA, while improving teacher self-efficacy and knowledge concerning PA implementation.