Despite the large volume of literature on child abuse and neglect generated over the past two decades, there has been relatively little comprehensive discussion of the limitations of different study designs for achieving major epidemiologic research goals. Although some improvements have been made in conventional research designs, there have been few efforts to develop new approaches. This article discusses the methodologic limitations of four observational study designs--ecologic, case-control, cross-sectional, and cohort--that dominate the child abuse and neglect literature; identifies key features of an "ideal" study of child maltreatment; and proposes a new mixed-design research strategy. The major advantage of the proposed strategy is that it greatly improves the ability to identify cases of child maltreatment from a well-defined population at risk representing diverse segments of the US population.