Previous tests of how people's valuations of safety vary with the level of baseline risk have left policymakers with a rather mixed message. Some data support the conventional hypothesis derived from economic theory that marginal valuations of risk changes increase with the baseline level, but other data reject it. With these indeterminate findings in mind, the present study adopted an in-depth quantitative – qualitative methodology to explore the preferences of a general population sample (N= 147) for safety programs targeted at hazards with different age groups at risk and numbers of deaths per year. The data clearly showed that self- or household interest mattered to people (e.g., programs in which the respondent or their household members were in the at-risk age group were evaluated more positively). The number of deaths also mattered, although not always for reasons strictly consistent with the conventional hypothesis. Rather, the qualitative findings suggested that evaluations may be driven by auxiliary assumptions about aspects of the programs induced by this information (e.g., people believe that a higher number of deaths is an indicator that more people are likely to be exposed to that hazard). These findings bring into question the extent to which quantitative responses from stated preference surveys can be taken at face value to form a reliable basis for public policy.