The simulation model for grassland, ModVege, was employed to examine grass growth dynamics and yield variability across two contrasting regions in Western Europe, Luxembourg and Switzerland. The model was calibrated with detailed field measurements from long-term growth monitoring programs in the two countries. To enable more realistic growth predictions under prolonged water stress, ModVege was adapted to better simulate post-drought recovery by integrating a minimum biomass threshold. By analysing twenty years of retrospective data, calibrated with detailed field measurements, we were able to identify significant regional differences: sites in Switzerland demonstrated higher, more stable yields, while those in Luxembourg showed greater inter-annual variability, that was mainly driven by differences in the water availability during the summer season. The climatic water balance was identified as strong predictor of annual yields, with site-specific correlations ranging from r2 = 0.50 to 0.85. Our findings emphasize the importance of site-specific management practices to improve vegetation resilience and yield stability in grasslands. As climate change intensifies drought frequency and alters precipitation patterns, these insights provide a solid foundation for adapting grassland management to ensure both productivity and long-term sustainability.