The account of cognitive vulnerability to depression offered by Beck's cognitive model is summarised. As this account currently lacks consistent empirical support, an alternative, related, account is presented. This proposes that, once a person is initially depressed, an important factor that determines whether their depression remains mild or transient, or becomes more severe and persistent, is the nature of the negative cognitive processes and constructs that become active and accessible in the depressed state. These interact with the nature of environmental difficulties, available social support, and biological factors, to determine whether a depression-maintaining cognitive-affective vicious cycle will be set up. Results from studies specifically designed to test predictions from this account have yielded positive results. Findings consistent with the hypothesis have also been obtained in other prospective studies which have shown that cognitive measures, administered in the depressed state, predict the future course of depression independently of initial levels of depression. The hypothesis is elaborated to incorporate the demonstrated relationship of elevated neuroticism to risk and persistence of depression. Recent views on the nature of sex differences in rates of depression, and on the relationship of attributional style to depression are also compatible with the hypothesis. It is concluded that the hypothesis has encouraging preliminary support.