Abstract: Using 2012–2022 data from the China Family Panel Studies (CFPS), four three-wave cohorts were constructed, with a primary sample of 1,358 adolescents and their parents. Parallel latent growth models for fathers, mothers, and adolescents decomposed body mass index (BMI) into intercept and linear change factors; robust maximum likelihood estimation with family-cluster correction was applied. Models 1 and 2 used 1,358 observations, whereas the fully adjusted Model 3 included 1,307 complete cases after adjustment for cohort-specific T0 per-capita household income and parental education. In Model 3, paternal and maternal BMI intercept factors were positively associated with the adolescent BMI intercept factor (B=0.200 and 0.221, respectively; both P<0.001). The paternal–adolescent slope association was not statistically significant (B=0.055, P=0.546), whereas the maternal–adolescent slope association was positive (B=0.196, P=0.008). A Wald test did not support a difference between the paternal and maternal slope paths (χ²(1)=1.160, P=0.282). Sensitivity analyses using a freely estimated time-loading growth model and WHO 2007 BMI-for-age z scores showed that the main association pattern was generally stable. Intergenerational BMI associations were most consistent for baseline levels; the significant maternal slope path should not be interpreted as statistically stronger than the paternal path.