Rainfall is one of the most important climatic elements affecting the water resources system, which is widely used in hydrological studies and management of water-related systems. The lack of or insufficient rainfall data is one of the main problems in design of water structures and management planning. In this study, the maximum 24-hour rainfall of the Urmia Lake Basin (ULB) for 63 rainfall selected stations for the period 30 years (1979-2008) was used. For this purpose, seven groups of attributes including climatic, geographical, statistical attributes, and their combinations along with five defined weighting scenarios using principal component analysis (PCA) and correlation coefficients matrix methods were applied. In general, 35 different pooling groups for Ward clustering method and 175 modes for ROI method were investigated. The results showed that each of the seven defined groups of attributes has different performance in terms of number of clusters, scattering of stations, and spatial pattern of regions. Furthermore, the results of weighting scenarios indicate that the performance of weighted regionalization models is better than non-weighted models in the estimation of extreme quantiles based one different attributes and defined weights, especially for Ward clustering method. Through the two regionalization methods, Ward and ROI clustering methods, the ROI approach demonstrated the much better performance and it can certainly be a very suitable and more effective alternative for the clustering method. In the ROI approach, the estimation results of the extreme amounts of maximum 24-hour rainfall were obtained with more accuracy, higher reliability, and less error. Additionally, improvement of skewness values for stations with short length of data indicated that the length of rainfall data can be less in comparison with length of flood data in the estimation of extreme values with the same accuracy; so that, the current study led to present a new formula. One of the important obtained results was also the estimation of the extreme values for the lack of data area. The results showed that applying the weight of attributes, especially the weight of each station, led to form a strong relationship in the region of interest and possibility of extreme values estimation for the lack of data regions that the only geographical attributes are available, which is one of the important results of this research. Keywords: Frequency analysis; Regional of influence; Ward clustering; Principal components analysis; correlation coefficient; regionalization; Urmia Lake basin.