Tree radial growth is the primary pathway for woody biomass accumulation and a key process in forest carbon sink formation. Understanding its response to climate variability is essential for improving forest carbon stock assessments and carbon cycle modeling under future climate change. Here, forest growth across North America were examined by integrating climate data and satellite-derived gross primary productivity (GPP) with tree-ring records. GPP was used to represent carbon assimilation and to assess how it mediates climate effects on radial growth at large spatial scales. Piecewise Structural Equation Modeling (pSEM) was applied to climate zone-- specific models to assess the mechanisms regulating tree radial growth at the regional scale. Results showed a widespread interannual lagged relationship between GPP and the ring-width index (RWI), with clear differences among climate zones. Lag effects were strongest in arid regions, moderate and stable in temperate regions, and mainly linked to climatic conditions during the previous cold season in boreal regions. pSEM analyses indicated good model performance across all zones. Interannual variability in GPP consistently positively linked radial growth, whereas long-term trend effects were weak and mostly negative. Radial growth was directly constrained primarily by moisture availability, whereas temperature-related variables mainly influenced growth indirectly through their regulation of carbon assimilation. The strength of moisture effects decreased from arid to temperate to cold zones, and vapor pressure deficit generally constrained growth. Overall, tree radial growth is a climate-regulated process characterized by strong regional heterogeneity and temporal legacy effects. Integrating carbon source-sink dynamics with long-term monitoring is essential for improving forest carbon cycle modeling and climate-adaptive forest management.