Although feedback has been identified as a key instructional feature in simulation based medical education (SBME), we remain uncertain as to the magnitude of its effectiveness and the mechanisms by which it may be effective. We employed a meta-analysis and critical narrative synthesis to examine the effectiveness of feedback for SBME procedural skills training and to examine how it works in this context. Our results demonstrate that feedback is moderately effective during procedural skills training in SBME, with a pooled effect size favoring feedback for skill outcomes of 0.74 (95 % CI 0.38-1.09; p < .001). Terminal feedback appears more effective than concurrent feedback for novice learners' skill retention. Multiple sources of feedback, including instructor feedback, lead to short-term performance gains although data on long-term effects is lacking. The mechanism by which feedback may be operating is consistent with the guidance hypothesis, with more research needed to examine other mechanisms such as cognitive load theory and social development theory.