The COVID-19 pandemic has been accompanied by the spread of misinformation and increasing polarization around public health measures such as social-distancing and national lockdowns. Previous research has suggested that citizens’ ability to determine the uncertainty of their knowledge – that is to “know what they don’t know” - is substantially impaired in areas marked by misinformation. In this study we investigated whether a similar phenomenon shaped public knowledge of COVID-19-relevant information in a national sample of Great Britain population. We used a signal-detection theory approach to measure metacognitive sensitivity; the extent to which one’s feeling of knowing matches the accuracy of one’s knowledge, and compare it to areas of knowledge that are less affected by misleading information. At the population level, metacognitive insight around COVID-19 knowledge appeared to be impaired compared to that in knowledge of general science. Crucially, individual differences in metacognitive insight into COVID-19 knowledge were predictive of health protective behaviours, vaccine intentions and attitudes toward public health measures, even after taking into account the level of knowledge itself and covariates such as education, income, political alignment. These results reveal the extent of the impact of misinformation on public beliefs, and establish confidence in accurate knowledge as an important target for science communication to promote compliance with public health and social measures.