Increased water demands, limitations of surface water resources, and excessive groundwater extraction have caused irreparable damage to the environment and natural resources. One of the major water quality issues in agriculture is nitrate toxicity. The Mashhad plain aquifer is one of the most important aquifers for water supply in Razavi Khorasan Province, Iran. This study aimed to investigate nitrate variations with some groundwater quality characteristics such as sodium, chloride, and electrical conductivity using geostatistical methods and to determine groundwater pollution trends in the Mashhad plain. Additionally, artificial intelligence models were employed to estimate nitrate concentrations. The latest nitrate measurements from 246 wells during 2006-2008 formed the basis for spatial and temporal analysis. Cross-validation techniques with evaluation criteria including Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) were used to compare methods. Results showed that nitrate concentration exhibited moderate spatial correlation in the plain. Cross-validation results indicated that among the methods used, log-normal kriging with minimum RMSE and MAE had higher accuracy, although the difference with ordinary kriging was negligible. Indicator kriging had the lowest accuracy in estimating groundwater nitrate levels. In simulating nitrate levels using artificial intelligence models with sodium, chloride, and electrical conductivity as inputs, neural networks demonstrated superiority over neuro-fuzzy systems.