Background The COVID-19 pandemic received widespread media coverage due to its novelty, an early lack of data, and the rapid rise in deaths and cases. This excessive coverage created a secondary “infodemic” that was considered to be a serious public and mental health problem by the World Health Organization and the international scientific community. The infodemic particularly affected older individuals, specifically those who are vulnerable to misinformation due to political positions, low interpretive and critical analysis capacity, and limited technical-scientific knowledge. Thus, it is important to understand older people’s reaction to COVID-19 information disseminated by the media and the effect on their lives and mental health. Objective We aimed to describe the profile of exposure to COVID-19 information among older Brazilian individuals and the impact on their mental health, perceived stress, and the presence of generalized anxiety disorder (GAD). Methods This cross-sectional, exploratory study surveyed 3307 older Brazilians via the web, social networks, and email between July 2020 and March 2021. Descriptive analysis and bivariate analysis were performed to estimate associations of interest. Results Major proportions of the 3307 participants were aged 60 to 64 years (n=1285, 38.9%), female (n=2250, 68.4%), and married (n=1835, 55.5%) and self-identified as White (n=2364, 71.5%). Only 295 (8.9%) had never started or completed a basic education. COVID-19 information was mainly accessed on television (n=2680, 81.1%) and social networks (n=1943, 58.8%). Television exposure was ≥3 hours in 1301 (39.3%) participants, social network use was 2 to 5 hours in 1084 (32.8%) participants, and radio exposure was ≥1 hour in 1223 (37%) participants. Frequency of exposure to social networks was significantly associated with perceived stress (P=.04) and GAD (P=.01). A Bonferroni post hoc test revealed significantly different perceived stress in participants who were exposed to social networks for 1 hour (P=.04) and those who had no exposure (P=.04). A crude linear regression showed that “some” social media use (P=.02) and 1 hour of exposure to social media (P<.001) were associated with perceived stress. Adjusting for sociodemographic variables revealed no associations with this outcome variable. In a crude logistic regression, some social media use (P<.001) and 2 to 5 hours of exposure to social media (P=.03) were associated with GAD. Adjusting for the indicated variables showed that some social network use (P<.001) and 1 hour (P=.04) and 2 to 5 hours (P=.03) of exposure to social media were associated with GAD. Conclusions Older people, especially women, were often exposed to COVID-19–related information through television and social networks; this affected their mental health, specifically GAD and stress. Thus, the impact of the infodemic should be considered during anamnesis for older people, so that they can share their feelings about it and receive appropriate psychosocial care.