Based on the series of daily SARS incident numbers from April 21 to May 20,2003,in Beijing area, the vacillating measurements affected by meteorological condition are separated from total tendency, with orthogonal polynomial methods. The results of correlation analysis show that they are significantly related with daily maximum temperature, daily range and relative humidity ahead of 8 to 9 days. On this basis, the regression estimation model can perfectly fit historic series of daily incidence fluctuation of this disease.