Using the daily gas load data and corresponding meteorological observational datasets in Hangzhou from 2008 to 2011, the variation characteristics of daily gas load and the impacts of meteorological conditions on gas loads are analyzed. Based on this, the daily gas load is predicted with the Elman network. The results show that the gas load increased significantly from 2008 to 2011 in Hangzhou. The peak point appeared usually one month before the Spring Festival, and the variation of the gas load decreased obviously during Spring Festival holidays. Temperature and pressure were the main factors responsible for the variation of gas load, with the correlation being most significant in winter. The gas load was significantly correlated to temperature negatively and pressure positively, most sensitive to the average air temperature from 6 ℃ to 15 ℃. According to the influence of meteorological factors and the Spring Festival, the daily gas load forecasting model in the wintertime is established with the Elman network. Based on the indexes such as mean relative error and correlation coefficient, it is illustrated that the Elman network has a satisfactory precision. While there appears noticeable fluctuations in the gas load, the result of simulation lags behind the observation.