U.S. regulatory agencies use the linear no-threshold (LNT) theory as a basis for risk estimation and regulatory decision making. In approaching risk assessment in this way, two problems immediately emerge. First, there is no generally agreed-upon principles that can be used to select one predictive theory to the exclusion of other biologically plausible alternatives. 1 As pointed out in Chapter 2 and later in this chapter, theory selection is driven by the need for simplicity and for conservative risk predictions. However, the simplest and most conservative theory may not necessarily be the most desirable because of the limited range of predictions that can be made. Complex theories with several parameters offer a wider range of predictions. 2 Second, theoretical risk predictions at environmental or occupational exposure levels preclude meaningful decision making because risk uncertainties are so large. The problem of uncertainty in decision making is discussed in Chapter 4.