During the COVID-19 pandemic, citizens were exposed to a wealth of information regarding the public health care system. Much of the provided information was numerical, such as number of deaths, infection rates, and number of people in intensive care. Despite this information can be expressed in various equivalent ways, it should not affect risk perception and decision making. To test this hypothesis, the present psychophysiological study aimed to investigate whether equivalent numerical formats such as raw ratio, absolute frequency, 1-in-X ratio, percentage, and probability-influence risk perception and affective response toward COVID-19. We measured risk perception using the Tripartite Model of Risk Perception scale and emotional response using the Positive and Negative Affect Schedule (PANAS) scale, along with psychophysiological measures (i.e., electrodermal activity [EDA] and facial expression). Self-protective behaviors were also measured in order to understand if a change in risk perception and emotional response leads to different levels of engagement in these kinds of behaviors. The reported findings indicate that the probability and percentage formats have less impact than the frequency formats. The latter has an impact on affective and experiential risk perception, negative emotions, and EDA. Theoretical and practical implications are discussed.