The present longitudinal study examined the cross-lagged relations between parental causal attributions of children's math success to children's ability, parental help, children's math performance and task persistence. A total of 735 children, their mothers, fathers and teachers were assessed twice--at the end of the second and the third grades. Children were tested in math and parents filled out questionnaires measuring causal attributions. Children's task persistence was reported both by parents and by class teachers. The results of path analyses indicated mutual negative effects between mothers' and fathers' help attributions and children's math performance, and a positive effect of mothers' ability attributions on children's math performance. Cross-lagged relations were also found for children's math performance and task persistence. However, relations between parental attributions and children's task persistence were evident only for fathers' ability attributions. The findings emphasise the importance of examining both mothers and fathers, home and school context.