The paper select 17 residential characteristics in Xi'an Qu Jiang New District and construct residential Hedonic model based on linear, logarithmic function form. Using of 169 group samples and 18 residential quarters of data, we analysis the characteristics of the price of residential features and price elasticity based on multiple regression methods. The results show that the logarithmic model has a higher fitting degree. Six variables enter into the logarithmic model including construction area, away from the South Lake distance, from the First Middle School of Qujiang, the distance from the San Zhao Cemetery, greening rate, the floor, while Seven variables enter into the linear model including construction area, away from the South Lake distance, the distance from the CBD, the floor, education facilities, etc.