Spatial variation of soil characteristics and vegetation conditions are viewed as the most important indicators of crop yield status. Therefore, this study was designed to develop a crop yield prediction model through spatial autocorrelation between the actual yield of potato (Solanum tuberosum L.) crop and selected yield status indicators (soil N, EC, pH, texture and vegetation condition), where the vegetation condition was represented by the cumulative normalized difference vegetation index (CNDVI). The study was conducted, during the period from December 2016 to March 2017, on a 30 ha center-pivot irrigated field located in Wadi-Ad-Dawasir area, Riyadh Province, Saudi Arabia. GPS assisted soil samples and yield data were collected from 120 sampling locations and analyzed for soil EC, N, pH and texture. The CNDVI; however, was derived from Sentinel-2A satellite images. Bivariate frequency ratio statistical approach was employed to assess the spatial correlation between crop yield and both of soil parameters and CNDVI. The spatial autocorrelation results were interpreted and further utilized for the generation of potato high yield susceptibility map. The accuracy of the obtained susceptibility (prediction) maps was investigated using the area under the curve (AUC) spatial validation technique. Soil pH was the most predictive factor in high yield zones, followed by texture, N, CNDVI and EC with prediction rates (PR) of 5.0, 4.2, 2.5, 1.9 and 1.0, respectively. The generated high susceptibility map represented the actual yield map with an AUC of 90%.