This paper investigates the role of specific and general metacognitive ability on specific and general academic achievement, controlling for the effects of intelligence. Four hypotheses were elaborated and empirically tested through structural equation modelling. The sample was composed by 684 students (6th to 12th graders) from a private Brazilian school, which answered to three intelligence tests and three metacognitive tests. The modeled hypotheses presented a good data-fit (χ2 = 51.18; df = 19; CFI = 1.00; RMSEA = 0.05), showing that the general metacognitive ability explained general academic achievement rather than intelligence, but did not explain specific academic achievement. On the other hand, specific metacognitive ability explained specific academic achievement rather than intelligence, but did not explain general academic achievement. The predictive power of the general metacognitive ability was greater than fluid intelligence in the explanation of general academic achievement. In the same line, specific metacognitive ability had a greater predictive power than intelligence and specific knowledge in the explanation of specific academic achievement. Finally, a new structural model of metacognition and its role in academic achievement are proposed.