In the post-truth era, particularly during the COVID-19 pandemic, an effective correction on misinformation is necessary to promote personal and public health. To better understand the effect of “correcting” misinformation, therefore, we investigated correction from different users on social media (e.g., individual users, fact-checking websites, and health organizations) and the frequency of correction (e.g., once vs. twice) in three online experiments. In each experiment, we evaluated participants’ perceived accuracy and willingness to share in terms of real and fake news of COVID-19, respectively. Across all experiments, a single correction from the health organizations effectively reduced participants’ perceived accuracy rating on the COVID-19 fake news. Experiments 2 and 3 revealed the effects of a single correction from individual users and fact-checking websites. Moreover, results of post-session questionnaires indicated that participants counted on the reliability of the sources in the correction. We did not obtain the consistent effects of frequent correction but verified the vulnerability of participants with high health anxiety to the COVID-19 fake news across all experiments. Overall, our study highlights the effects of user-initiated correction regardless of whether the user is an individual or an organization, as long as the correction contains a reliable source.