The effect of dietary intake on body weight may vary based on individual genetic differences. However, children are rarely used in such investigations. The aim was to identify possible genetic moderation through polygenic scores (PGS) for BMI, of the association between dietary intakes and BMI in children. The study sample included children who were part of a French-Canadian birth-cohort study. BMI data was available on seven occasions between ages 4 and 13 years. FFQ (juice and fruit drinks, sweets and snack foods, meats, and fruits and vegetables) and 24-h dietary recall (proteins, lipids, carbohydrates, total energy) data were available up to 4 years. Linear mixed models were used to account for repeated BMI measurements. The consumption of juice and fruit drinks (in girls), sweets and snack foods, fruits and vegetables, proteins, lipids, carbohydrates and total energy were associated with BMI. Associations with BMI increased with age (kg/m2 per year) for fruits and vegetables (β: −0.03, 95%CI: −0.06;−0.01), lipids (β: 0.11, 95%CI: 0.01;0.22), carbohydrates (β: 0.05, 95%CI: 0.01;0.08), and total energy (β: 0.07, 95%CI: 0.02;0.12), and with higher values of a PGS (kg/m2 per SD) for proteins (β: 0.54, 95%CI: 0.03;1.06), lipids (β: 0.63, 95%CI: 0.12;1.13), and total energy (β: 0.32, 95%CI: 0.06;0.58). Using longitudinal data, we showed that the associations between specific dietary intakes and BMI may vary depending on age and genetic susceptibility in childhood.