Metacognition is typically measured by collecting self-reported information from participants while they complete a cognitive task. Recent evidence suggests that eliciting such metacognitive information from participants can impact both their metacognitive processes and their cognitive performance. Although there are contradictory findings regarding the magnitude and even the direction of this effect, recent evidence has converged to provide a clearer picture of the mechanisms that determine reactivity. Here, we provide a review of the evidence that measures of metacognition, namely think-aloud protocols, judgments of learning, and confidence ratings, are reactive. We argue that reactivity has important implications not just for the measurement of metacognition, but for metacognition theorizing because reactivity can provide insights into the cues participants use to monitor their performance. Drawing from this synthesis of evidence, we propose a framework for studying reactivity that integrates cue processing accounts of reactivity with existing models of metacognition. We conclude the review by addressing some of the pertinent questions yet to be comprehensively addressed by reactivity research, including how researchers should best address issues of reactivity when using experimental designs.