This paper is concerned with the kind of work that is needed in order to provide a more solid scientific foundation for the belief that life stress causes illness. Although indirect evidence from controlled laboratory studies and from extreme situations provides strong indications of the general relationship between stress and illness, the evidence concerning the effects of naturally occurring and more ordinary stressful life events is less clear. Although we have a body of research indicating that life events are associated with a wide range of disorders and distress, it does not provide a clear picture of the nature and strength of this relationship. In order to clarify our understanding of how stressful life events affect health and illness, we will need to deal with methodological issues that have emerged from research to date and to tackle a major substantive problem that has been largely neglected. The methodological issues are three: a) definition of the populations of life events to be studied; b) measurement of the magnitudes of the life events; and c) use of a research design appropriate to the question to be answered. The substantive issue concerns factors that mediate the impact of life events. Each of these issues is examined in turn in this paper.