This study investigates math anxiety (MA) by comparing trait-components of MA with real-time assessments of situational anxiety responses (state-components) in children. The research to date on MA in children is somewhat disparate in regard to methodology, and firm conclusions regarding the relation of MA to intelligence, math achievement and learning motivation are not readily drawn. Typically, the measures used in the MA research have differed by operationalizing either trait-MA and/or state- (or statelike)-MA, but have failed to compare the implications of their respective assumptions and the significance of their findings. Trait-MA and state-MA, self-ratings of math skills, attitudes towards mathematics, math achievement, the social anxiety, test anxiety, learning motivation and intelligence of 1,179 students (48.1% girls) from grades 4 and 5, were assessed. The findings yield evidence of a pronounced state-trait discrepancy. A negative correlation between state-MA and math achievement was observed for all intelligence levels, even when controlling for test- and social-anxiety traits, while there was no negative relation between trait-MA and achievement. State-MA was associated with lower intelligence, lower self-ratings, more negative attitudes, higher performance avoidance and work avoidance goals. In contrast, trait-MA was slightly related to higher mastery approach goals. The failure to adequately differentiate between state- and trait-based research into MA appears to be one reason for key inconsistencies between research findings and warrants further investigations.