Background:The COVID-19 pandemic has increased risks to people's health due to restrictions on movement and access to exercise facilities. This study compares and analyzes perceptions and key attributes of exercise among Korean men and women in the context of the COVID-19 pandemic by using data collected from social media. Methods: Data were collected from January 2020 to June 2021. The search keywords were classified into "COVID-19, men, and exercise" and "COVID-19, women, and exercise", and data were collected through Textom (The Imc Inc., Daegu, Republic of Korea), a big data collection and analysis platform. Frequency analysis, TF-IDF analysis, and degree centrality analysis were performed using Textom. In addition, a CONCOR (CONvergent and CORrelation) analysis was performed using UCINET 6 (Analytic Technologies Corp., Lexington, KY, USA) to visualize related words and analyze clusters. Results: In total, 10,703 data points of 5.935 MB were collected and analyzed. The top 50 words related to men's and women's exercises of COVID-19 were derived through frequency analysis, TF-IDF analysis, and degree centrality analysis. We also derived four common clusters (exercise method, COVID-19 situation, friends, and sports). In addition, three clusters (women's fitness, children and hashtags) for men's exercise and two clusters (weight loss and Olympics) for women's exercise were derived separately. Conclusions: The study results show that it is possible to provide suggestions and health improvement plans for exercise based on gender. In addition, our findings can serve as an academic basis for developing health promotion programs related to exercise during the COVID-19 pandemic, which can contribute to reducing the economic costs and health risks associated with the pandemic.