Attributional style models of depression in adults (Abramson et al. 1989, 1978) have been adapted for use with children; however, most applications do not consider that children's understanding of causal relations may be qualitatively different from that of adults. If children's causal attributions depend on children's level of cognitive development, then support for attributional models of depression in young people will vary with cognitive development. In this paper, a new measure of cognitive development, the Peabody Causal Reasoning Test (PCRT), is introduced to assess children's understanding of ability versus effort, task difficulty, and luck as causal factors. Analyses revealed that in 8- to 16-year-old children, failure to control for level of cognitive development suppressed empirical support for cognitive diathesis-stress models of depression. Statistically controlling for measures of cognitive development strengthened support for this model.