Some twin studies suggest that substance initiation and dependence are part of a complex, two-stage process and that some genetic influences are stage-specific, acting on either the transition from abstinence to initiation, or on the transition from use to dependence. However, questions remain about the two-stage model, especially for illicit drugs. Using a familial aggregation design, we tested the hypothesized two-stage model of dependence on illicit substances and alcohol in a large, nationally representative sample. Family history of drug or alcohol problems is significantly associated with initiation that does not progress to dependence (i.e., conditional initiation). Furthermore, family history of drug or alcohol problems is significantly associated with dependence even after conditioning on factors influencing initiation (i.e., conditional dependence). These results suggest that substance initiation and dependence involve at least partially distinct familial factors. The possibility that different genetic factors affect initiation and dependence has important implications for control group selection in case-control genetic association studies, and may explain some inconsistent results for drug dependence. If some genetic factors are stage-specific (i.e., not common across initiation and dependence), inclusion of abstainers in the control group may mix the genetic effects for initiation with those for transition to dependence, providing unclear results. Depending on the specific question about the nature of the genetic effect (whether on initiation, on dependence, or both), investigators designing case-control genetic association studies should carefully consider inclusion and exclusion criteria of the control group.