Groundwater is one the major sources of natural water being exploited excessively for various uses in India. Thus, it is very essential to monitor the spatial and temporal variability of groundwater quality. Geo-Statistical Interpolation using GIS has been considered as the best and most advanced method for the interpolation and prediction studies of groundwater pollution and quality, and is adopted universally. In this paper, ordinary Kriging with logarithmic data transformation has been used to interpolate and predict the spatial variation of groundwater quality parameters - EC, TDS, pH, Na+, Ca2+, Bi-Carbonate, Fluoride, Chloride, Sulphate and Nitrate using data pertaining to 48 well locations in the Dungarpur tehsil. Data was transformed and normalized using Logarithmic Transformation Method and Semivariograms were drawn and analyzed for selecting the suitable model. The best Semivariogram model was obtained based upon cross validation and on the lesser RMSE criterion and Coefficient of Determination. The results show that the best semivariogram model based on RMSE varied for each water quality parameter. For log transformed data Exponential model was found suitable for EC, TDS, Na+, TH etc.;  Spherical model for Ca2+ ; Chloride Gaussian Model for Chloride. For original or raw for non-transformed data Exponential Model was found suitable for Fluoride, Sulphate and Nitrate; and Gaussian Model for pH and Bi-Carbonates.