SYNOPSIS This paper discusses the magnitude of the effect of life events in the causation of psychiatric illness. It is argued that an established epidemiological concept, relative risk, provides a useful measure of association which can be approximately adapted for retrospective controlled studies. Examination of studies employing general population controls consistently indicates effects of some importance, with risks of illness increased by factors of between 2 and 7 in the 6 months after an event. Risks are greater for the more stressful types of events, greater for depression and neuroses than schizophrenia, and even greater for suicide attempts. However, similar events occur commonly and a large proportion of event occurrences are not followed by illness. Events must interact with a wide variety of background factors, and the appropriate model is one of multifactorial causation.