Human suicide presents a fundamental problem for the scientific analysis of behavior. This problem has been neither appreciated nor confronted by research and theory. Almost all other behavior exhibited by humans and nonhumans can be viewed as supporting the behaving organism's biological fitness and advancing the welfare of its genes. Yet suicide acts against these ends, and does so more directly and unequivocally than any other form of maladaptive behavior. Four heuristic models are presented here to account for suicide in an evolutionary and sociobiological framework. The first model attributes suicide to the extraordinary development of learning and cultural evolution in the human species. Learning may make human behavior so independent of biological constraints that it can occasionally assume a form entirely contrary to the principles of biological evolution. The second model attributes suicide to a breakdown of adaptive mechanisms in extremely stressful novel environments. The third model involves kin and group selection, arguing that in limited circumstances suicide may occur because of beneficial effects it has on other, surviving individuals who share the suicidal individual's genes. The last model suggests that suicide should be tolerated by evolution when it has no effect on the gene pool. This model holds particular promise in accounting for aspects of suicide not attributable to culture. The evidence indicates that suicide is most common in individuals who are unlikely to reproduce and unable to engage in productive activity; such individuals are least capable of promoting their genes. A complete explanation of suicide may derive only from an analysis of its biological significance.