The effect of deficiencies in social relationships has been studied prospectively in a community sample. In the half who were exposed to the higher level of adversity, these deficiencies explained 30 per cent of the variance in neurotic symptoms four months later. They explained only 4 per cent in those with lower adversity. Contrary to expectation, it was not the lack of relationships, but perception of these as being inadequate under adversity, which had by far the stronger predictive power. This may mean that actual conditions in the immediate social environment are not important for neurosis. Instead, intrapsychic and personality factors may have a much stronger effect.