The present article studies the correlation between data of the BPR-5 Intelligence test with a school assessment of mathematics and portuguese language and a questionnaire on family and academic performance from a sample of 679 ninth grade students from four elementary private schools.The results indicate a strong correlation and statistically significant with IQ test scores (r =, 58, p <0.01), showing high loads on Fluid Intelligence (Gf). Some correlations between intelligence and family and school aspects were also found. A longitudinal analysis (5th to 9th grade) was applied using the Latent Growth Curve Model, which investigated the average initial variance (intercept) and the average growth (slope) in the subjects' academic performance (AP), in two models (with and without the independent variable BPR), aiming to investigate the predictive capacity of Gf in AP. When the variable BPR was inserted, its impact on the intercept was estimated at 20,288 and on the slope, 6,381. These outcomes indicate the increase on initial performance and growth in AP due to each additional point in the BPR score. The correlation between the intercept and the slope was negative and statistically significant (r=-224,156, p <0.01), indicating that the subjects who presented lower initial AP had a higher growth in the evaluated period. Thus, the hypothesis of Gf's predictive ability on AP was confirmed, which corroborates with the literature results.