<p>Himalayan rivers are crucial for freshwater supply, ecosystem services, and livelihoods across their course from mountains to floodplains, but climate change, human activities, and geological factors are increasingly degrading water quality. Given these challenges and the vulnerability of the Sikkim Himalayan ecosystem, this study provides a comprehensive assessment of water quality by employing hydrogeochemical analysis, entropy-weighted water quality index, and advanced machine learning models to develop an effective decision-support framework for water resource management. A total of 160 water samples were collected from households, springs, and rivers across different seasons, and it was found that spatial and seasonal variability significantly influence the key parameters. Carbonate and silicate weathering dominantly controlled the chemical constituents of water (60–80%), along with localised anthropogenic influences. Various water quality indices were applied to assess the suitability of water for agricultural and domestic uses, with majority of samples classified from excellent to suitable categories, however, certain locations, especially those in densely populated and near geothermal regions, exhibited poorer water quality. In addition, advanced algorithm-based models were employed to further forecast and validate water quality scenario. Among several tested models, the Ridge Regression demonstrates superior performance (R² = 1) with the lowest prediction error (RMSE = 0.02) in the region. SHAP sensitivity analysis revealed that NO₃⁻ and K⁺ are the key pollutant exerting the most significant influence on the water quality. This study supports in the formulation of sustainable water management practices aligned with SDG 6, enhancing the long-term resilience of Sikkim’s Himalayan water resources against climate change.</p>