A new instrument, the Why It Happened questionnaire, was developed for assessing characterological and behavior self-blame in two samples of third- through ninth-grade children (Sample 1: n = 121; sample 2: n = 322). Children's responses were very consistent within specific classes of negative events but only moderately consistent across time. Exploratory and confirmatory factor analyses of open-ended and self-rating responses, respectively, revealed that children's self-blame attributions were only moderately consistent across different classes of negative events, and that children's tendency to make characterological and behavioral attributions were unrelated for academic events and moderately correlated for social events. Multiple regression revealed that characterological self-blame about academic and social events were not significantly related to self-reported depression until ninth grade. Implications for cognitive diathesis-stress models of depression for children are discussed.