Many investigators have studied the relationship between life events and psychiatric illness to discover whether certain events predispose to some disorders or precipitate them, and to discover the effect of illnesses already in progress on subsequent life events and the effect of events on established illnesses. The validity of such studies depends, among other things, on the selection of suitable controls. For example, if adult patients with depression differ from a group of controls in that they had a higher incidence of parental death, it might indicate that bereavement had made them more susceptible to the development of the illness. But if the depressed patients were from a lower social class than their controls, the greater incidence of bereavement might be explained by the fact that there is a higher death rate in that stratum of society. This would then cast doubt on a conclusion about a causal relationship in that group between childhood bereavement and depression in adult life.