This chapter explores how the representation of statistical information affects the understanding of risks and uncertainties in medical contexts. Using mammography screening as a prime example, it is shown that problems in understanding and dealing with numbers are often due to poorly designed information environments, rather than to internal deficiencies of the human mind. For three types of statistical information that physicians and patients often encounter-conditional probabilities, single-event probabilities, and relative risks-a representation is proposed that facilitates understanding. These are compared to the representations actually used in published materials about mammography screening. Factors in the environment that can contribute to innumeracy are identified and the question of why risks are not always communicated in a transparent manner is addressed. Finally, recommendations are formulated for changes, in both the information environment and the institutional and legal environments, that could help foster statistical thinking and informed decisions about medical screening.