This study evaluates predictions central to the extended parallel process model (EPPM) in the context of dense breast notifications. Many EPPM propositions have gone untested and competing predictions to the model have been evaluated to an even lesser extent. Also left as an open question is exactly how perceived threat and efficacy constructs should be treated in health communication research. Using experimental data collected from women likely to receive dense breast notification letters (i.e., aged 40 to 50 years) in states with and without dense breast notification legislation, this study explicitly tests EPPM predictions regarding danger control responses. These data were largely unsupportive of the EPPM's predictions and instead finds that negative affect is more of a direct predictor of intention than expected. These data also provide evidence supporting the separate treatment of the perceived severity, susceptibility, self-efficacy, and response efficacy variables, contrary to convention in EPPM research. Implications for breast density research and EPPM theorizing are discussed in light of these findings.