Differentials in health and longevity by socioeconomic status and by the nature of social relationships have been found in innumerable studies in the social and medical sciences. Three categories of explanations for the observed patterns have been proposed: causal mechanisms through which the social environment affects health status or the risk of dying; selection or reverse causal pathways whereby a person's health status affects their social position; and artifactual mechanisms, such as measurement error. The general consensus among researchers is that the observed disparities in health are driven largely by a complex set of causal processes rather than by selection or by artifactual mechanisms. This paper explores the set of arguments and strategies that researchers have used to arrive at this conclusion. As part of this undertaking, we assess whether inferences regarding the minor contribution of selection to the overall association between social factors and health are justifiable. In addition, we identify current avenues of research that are providing new insights into the causal pathways linking social factors and health.