An estimated 47.8 million Americans get foodborne illnesses each year. The risks posed by foodborne contaminants would be reduced, if consumers routinely practiced the handling practices recommended by food safety educators. However, despite many educational efforts, they typically do not follow these practices. As a result, regulatory agencies' episodic instructions, regarding individual outbreaks, have a vital role in consumer protections. They alert consumers about situations where immediate action is needed, instruct them in what to do, and reassure them when their food is not suspect. To improve the effectiveness of foodborne outbreak communications, we applied a mental models approach seeking to identify the correct beliefs, knowledge gaps, and misconceptions that guide consumers’ responses to food recalls. It began by creating an expert model, summarizing the major factors known to affect risk levels. It proceeded to in-depth open-ended mental models interviews, structured about the expert model. The interviews were conducted with thirty consumers, recruited from Pittsburgh area. It was a diverse group, varying along dimensions that might affect food safety beliefs. We found that food recalls increase the trust of the respondents in the U.S. food safety system because the necessary steps are taken to make sure that people are safe. Although most recalls never officially end, the interviewees incorrectly believed that they will be informed, when a recall is over. Exaggerating their ability to control risks, they believed that consuming recalled products would depend on issues like whether they could eliminate the risk by themselves (e.g., washing and cooking). Their self-diagnoses of foodborne illnesses demonstrated shorter incubation and recovery periods suggesting significant misperceptions. The results provide the foundation for structured surveys, suited to evaluating the population prevalence of the most important beliefs and for designing communications to improve these mental models and evaluating their effectiveness.