Public policies are typically established to eliminate important social problems (e.g., minority discrimination, crime, poverty). And the importance of these problems, and urgency people feel about addressing them, is influenced by perceptions of their prevalence. These perceptions, however, can be unwittingly biased by extraneous sources of information that lead some either to overestimate or underestimate the seriousness of the problem at hand. We review empirical work on the construction of perceptions of frequency and representativeness and the processes that underlie them, and show that these perceptions are often biased in ways that differ over segments of the population. The implications of these findings for developing public policy initiatives and de-biasing strategies are discussed.