Abstract Background We aimed to identify, synthesise and evaluate randomised control trial evidence on the effects of healthcare professional training on the delivery quality of health behaviour change interventions and, subsequently, on patient health behaviours. Methods Systematic review with narrative synthesis of effects on delivery quality and meta-analysis of health behaviour outcomes. We searched: Medline, EMBASE, PsychInfo, AMED, CINAHL Plus and the Cochrane Central Register of Control Trials up to March 2019. Studies were included if they were in English and included intervention delivery quality as an outcome. The systematic review was registered on PROSPERO (registration: CRD42019124502). Results Twelve-studies were identified as suitable for inclusion. All studies were judged as being high risk of bias with respect to training quality outcomes. However with respect to behavioural outcomes, only two of the six studies included in the meta-analysis had a high risk and four had some concerns. Educational elements (e.g. presentations) were used in all studies and nine included additional practical learning tasks. In eight studies reporting delivery quality, 54% of healthcare professional communication outcomes and 55% of content delivery outcomes improved in the intervention arm compared to controls. Training that included both educational and practical elements tended to be more effective. Meta-analysis of patient health behavioural outcomes in six-studies found significant improvements (Standardised mean difference (SMD): 0.20, 95% confidence interval: 0.11 to 0.28, P < 0.0001, I² = 0%). No significant difference was found between short (≤6-months) and long-term (> 6-months) outcomes (SMD: 0.25 vs 0.15; P = 0.31). Conclusions Delivery quality of health behaviour change interventions appears to improve following training and consequently to improve health behaviours. Future studies should develop more concise /integrated measures of delivery quality and develop optimal methods of training delivery.