Leaf area index (LAI) regulates land–atmosphere exchanges and influences evapotranspiration (ET), seasonal water storage, and streamflow, but, in hydrological modelling, the value of constraining LAI with observations at the global scale remains poorly quantified at the basin scale. We assess the impact of assimilating satellite-derived LAI (GEOV2-AVHRR) in the global land data assimilation system LDAS-Monde coupled to ISBA-CTRIP over 1982–2017, evaluated against observation-based products and discharge records from 2139 gauged basins. Basins are grouped into hydro-biospheric regimes using observation-based descriptors to interpret regime-dependent responses. LAI assimilation improves vegetation climatology and phenology and yields spatially coherent gains in ET skill, alongside robust reductions in errors of snow water equivalent variations (ΔSWE) across regimes. Streamflow improvements are more modest and heterogeneous—consistent with unchanged precipitation forcing and compensating errors among flux and storage terms—but discharge biases are reduced in several regimes. Overall, LAI assimilation provides a practical pathway to improve key water-cycle components (ET and ΔSWE) and to deliver incremental, process-consistent benefits for basin-scale streamflow realism.