This study examined the structural relationships among cognitive constructs (intelligence and achievement) and affective constructs (perceived parental help, effort and self-concept). It was proposed that the relationships are not invariant across gender. The sample consisted of 219 boys and 133 girls from elementary and preparatory public schools in Al Ain in the United Arab Emirates. Intelligence (IQ) was measured by the Test of Non-verbal Intelligence (TONI) and parental help was measured by 4-Likert-type items. Effort was measured by 4-Likert-type items. Self-concept (SC) was measured by 8-Likert-type items taken from the SDQ I (Abu-Hilal, 2000). Mathematic Achievement was the scores of students in mathematics from school records. The structural model assumed that IQ would have an effect on parental help, effort, SC and achievement. Parental help would have an effect on effort, SC and achievement. Also, effort would have an effect on SC and achievement. Finally, SC would have an effect on achievement. The structural model was tested for invariance across gender. The measurement model proved to be invariant across gender and so was the structural model. The non-constrained model indicated that the structural relationships among the variables do vary according to gender. For example, boys benefited from parental help by exerting more effort while girls did not. Boys with high IQ exerted more effort than boys with low IQ; but girls with high IQ exerted the same amount of effort as girls with low IQ. The model explained 45% and 39% of the variance in math scores for boys and girls, respectively.