A prospective study of 400 largely working-class women with children living at home has used measures of self-esteem and 'social support' to predict the risk of depression in the following year once a stressor had occurred. Actual support received at the time of any crisis in the follow-up year was also measured. Self-esteem was correlated quite highly with some of the measures of support. A core tie was defined as a husband, lover or someone named as very close at first contact. Negative evaluation of self (i.e. low self-esteem), and various indices of lack of support from a core tie at the first interview, were associated with a greatly increased risk of subsequent depression once a stressor occurred. Lack of support from a core tie at the time of the crisis was particularly highly associated with an increased risk. There was also a high risk among those who were 'let down' - that is, for those who did not receive the support which they might have expected in terms of the first interview material. It is concluded that it is essential for prospective enquiries to take account of the actual mobilization of support in the follow-up period.