Prevention research in public health requires quality data. In injury prevention research, "official" data sources, such as medical or law enforcement data, often do not possess the required depth or completeness. Self-reported data can fill this gap. Such data allow us to understand knowledge, attitudes, exposures, and behaviors associated with injury risk. Self-reported data are also needed to understand outcomes that are often missing from official sources, such as victimization by an intimate partner that is not reported because of concerns about legal consequences and less severe injuries from suicide attempts that go untreated. Data on risk and protective factors and specific types of violence exposures can often only be obtained by directly asking those affected. In addition, "official" data sources are rarely representative. Random-digit-dialing (RDD) surveys are a method of obtaining representative self-reported data. The RDD approach is relatively cost effective, handles non-English-speaking households with relative ease, and possesses a well-developed theory for constructing sample weights. However, there are significant challenges to using RDD surveys. These include declining participation rates; possible self-selection bias, since potential respondents can choose to opt out of the survey; and, with sensitive topics such as intimate partner violence, the need to anticipate potential risks for participants. This theme issue provides suggestions for how we can improve the design and implementation of RDD surveys in a manner that is both practical and ethical.