Water quality (WQ) data are essential for water resources as well as WQ management. However, due to limited fiscal resources, in many cases, historical WQ data may not be available for the watershed(s) of interest. Recently, few artificial intelligent and parametric statistical approaches were proposed for the estimation of WQ characteristics at ungauged watersheds. In this study, the main goal was to develop nonparametric statistical approaches for the estimation of the three main quartiles of different WQ indicators in ungauged watersheds using watershed attributes as predictors. Four nonparametric approaches were based on the region of influence (ROI) and Theil-Sen nonparametric multiple regression (TSMR). The ROI was used to identify watersheds similar to the ungauged watershed, and TSMR was calibrated and validated to estimate WQ quartiles in ungauged watersheds. The main proposed approach combined the ROI and TSMR. In the second approach, the Spearman correlation analysis was introduced as a filter, to identify watershed attributes that are highly correlated with the WQ quartile, to improve the ROI performance, (CO)ROI-TSMR. In the third approach, a step forward selection technique was integrated with the TSMR to facilitate the selection of the model predictors (watershed attributes), ROI-(SF)TSMR. In the fourth approach, both the (CO)ROI and (SF)TSMR were used, (CO)ROI-(SF)TSMR. The four nonparametric statistical approaches were calibrated and validated, to estimate the three main quartiles for three different WQ indicators, using data from 50 watersheds in the Nile Delta, Egypt. A Jackknife validation procedure was applied to evaluate the performance of the four approaches. The results indicated that the (CO)ROI-(SF)TSMR approach outperformed the three other approaches.