This review investigated three research questions (i) What is the utility of social cognitive theory (SCT) to explain physical activity (PA)?; (ii) Is the effectiveness of SCT moderated by sample or methodological characteristics? and (iii) What is the frequency of significant associations between the core SCT constructs and PA? Ten electronic databases were searched with no date or sample restrictions. Forty-four studies were retrieved containing 55 SCT models of PA. Methodological quality was assessed using a standardized tool. A random-effects meta-analysis revealed that SCT accounted for 31% of the variance in PA. However, methodological quality was mostly poor for these models. Methodological quality and sample age moderated the PA effect size, with increases in both associated with greater variance explained. Although self-efficacy and goals were consistently associated with PA, outcome expectations and socio-structural factors were not. This review determined that SCT is a useful framework to explain PA behaviour. Higher quality models explained more PA variance, but overall methodological quality was poor. As such, high-quality studies examining the utility of SCT to explain PA are warranted.