The proposition that attributional style is a risk factor for depression, with people who make external, unstable attributions for good outcomes, and internal, stable attributions for bad outcomes being particularly vulnerable, was tested in a study of employed and unemployed youngsters. Among the former, greater self-esteem was associated with internal attributions for good outcomes, and less depressive affect was associated with internal, stable attributions for good outcomes. No such relationships were observed in the unemployed. By contrast, attributions for bad outcomes were related to both depressive affect and self-esteem in the unemployed, but were related only to depressive affect in the employed. In the unemployed, lower depressive affect and higher self-esteem were both associated with unstable attributions, and lower depressive affect was associated with external attributions. In the employed lower depressive affect was associated with external, unstable attributions. Although these relationships were generally consistent with the hypothesis, attributions made three years earlier when respondents were still at school were only weakly related to subsequent measures of psychological well-being. Moreover, many changed their attributions over time, a finding that casts doubt on the assumption that attributional style can be regarded as a stable characteristic in young people.