The main purpose of structural health monitoring (SHM) is to provide accurate and real-time information about the state of a structure, which can be used as ob-jective inputs for decision-making regarding its management. However, SHM and decision-making occur at various stages. SHM assesses the state of a struc-ture based on the acquisition and interpretation of data, which is usually provided by sensors. Conversely, decision-making helps us to identify the optimal man-agement action to undertake. Generally, the research community recognizes peo-ple tend to use irrational methods for their interpretation of monitoring data, in-stead of rational algorithms such as Bayesian inference. People use heuristics as efficient rules to simplify complex problems and overcome the limits in rationality and computation of the human brain. Even though the results are typically satis-factory, they can differ from those derived from a rational process. Many heuris-tic behaviors have been studied and demonstrated, with applications in various fields such as psychology, cognitive science, economics and finance, but not yet in SHM-based decision. SHM-based decision-making is particularly susceptible to the representativeness heuristic, where simplified rules for updating probabili-ties can distort the decision maker’s perception of risk. In this paper, we examine how representativeness affects the interpretation of data, providing a deeper un-derstanding of the differences between a heuristic method affected by cognitive biases and the classical approach. Our study is conducted both theoretically through comparison with formal Bayesian methods as well as empirically through the application to a real-life case study about the evaluation of a bridge safety.