Although substance-related disorders are heritable, the genetic factors contributing to vulnerability to these disorders are expected to be complex. Nonetheless, identifying genes underlying this vulnerability and understanding their relationship with environmental factors and behavior holds the promise of dramatic advances in diagnosis, prevention, and treatment. The search is complicated by a number of factors, however, including the weak validity of psychiatric diagnosis for identifying gene carriers, the complexity of the brain and behavior, and the numerous intervening variables between genetic transcription and its behavioral consequences. One strategy for bridging this theoretical gap is to study endophenotypes--biologic correlates of disorders that precede their overt development, may have higher reliability than behavioral measures, and present simpler relationships with a smaller number of genes. This article reviews research suggesting the usefulness of several putative endophenotypes for substance-related disorders, including 1) reduced P3 amplitude of the visual event-related potential, 2) increased EEG beta power, 3) a lowered level of response to an alcohol challenge, and 4) the inability to modulate autonomic nervous system reactivity under the stress of anticipating a predictable aversive stimulus.