<p>Groundwater quality assessment is necessary in agricultural regions where excessive extraction has occurred for decades, and where water stress is affected by poor water resource management and the effects of drought. The study was conducted in a region with a high presence of rainfed and irrigated agricultural activity that uses groundwater from the Celaya Valley aquifer, in the municipality of Guanajuato, Mexico. The analysis was performed through hydrogeochemical characterization, and water quality indices for agricultural use were evaluated. Forty-six groundwater samples were analysed to conduct a hydrogeochemical characterization, considering the primary ions. This allowed for the identification of salinity and sodicity hazards. Fourteen indicators used in assessing water quality for agricultural use were evaluated. To reduce dimensionality, principal component analysis was applied, identifying two principal components, which account for 83.31% of the total variance. Principal Component 1 (55.36%) represents salinity-sodicity processes, and Principal Component 2 (27.95%) shows water hardness processes. Using the unsupervised K-Means machine learning algorithm in the principal component space, three clusters were identified, corresponding to the following hydrogeochemical processes: Cluster 1 (medium salinity-sodicity) Cluster 2 (low salinity-sodicity), and Cluster 3 (high salinity-sodicity). Salinity and sodicity hazard assessments, using Electrical Conductivity, determined that 11% of the samples had a high salinity hazard. The Sodium Adsorption Ratio indicated that 13% of the samples had a medium to very high sodicity hazard. The integrated methodological framework combines hydrogeochemical analysis, supported by multivariate statistics, and the evaluation of water quality indices for agricultural use, providing optimal interpretation of groundwater quality conditions.</p>