Diabetic retinopathy is a leading cause of vision loss globally. The current diagnostic thresholds for diabetes are still based on historic data correlating glycemic parameters with retinopathy; however, an excess prevalence of retinopathy has also been reported in prediabetes. We aimed to determine the reported prevalence of retinopathy in adults with prediabetes. We performed searches using MEDLINE, EMBASE, PubMed, Web of Science, CINAHL, Google Scholar and the Cochrane databases from inception to August 1, 2020. We evaluated methodological quality and certainty of the evidence using a validated risk of bias tool and GRADE, respectively. Twenty-four studies (8,759 participants with prediabetes) were included after screening 5,155 abstracts and reviewing 98 full-text records. Nineteen studies (79%) reported population-based data. Retinopathy prevalence estimates ranged between 0.3-14.1% (median 7.1%, interquartile range 2.4-9.7%), with high variance in estimates due to differing screening methods, retinopathy grading protocols and study populations. We judged this as low-certainty evidence using GRADE, downgrading for risk of bias and inconsistency. From studies that compared both populations, post hoc analysis revealed a lower median retinopathy prevalence in normal glucose tolerance (3.2%, interquartile range 0.3-7.3%) than prediabetes (6.6%, interquartile range 1.9-9.8%). These data suggest an excess prevalence of retinopathy in prediabetes.