To determine the predictive role of metacognitive, cognitive, and affective factors in solving verbal math problems, a cluster sample of 450 junior high school  students was given a battery of tests and questionnaires measuring the students' ability in problem solving, predictive representation, meta-cognitive supervision, and intelligence, as well as their meta-cognitive knowledge and beliefs. The analyses of the collected data revealed that more than 41% of the variance in problem solving ability is explained by the variables included, with the metacognitive factors having the largest share, followed by the cognitive and affective factors.