The interdecadal variation of spring dust storm frequency (SDSF) in China and its possible affecting factors are examined on the basis of landsurface air temperature (SAT) and NCEP/NCAR reanalyzed datasets from 1954 to 2007. The results suggest that SDSF is negatively correlated with the SAT significantly around Lake Baikal, where the maximum correlation coefficient is -0.8. The warming around Lake Baikal can reduce the frequency of Mongolian cyclones and resulted in the decrease of SDSF in China over the past decades. It is found that the winter temperature at 850 hPa is positively correlated with the spring SAT significantly around Lake Baikal; therefore, a linear regression model between averaged winter air temperature index over the region of 80°〖KG-*9〗E to 140°〖KG-*9〗E, 40°〖KG-*9〗N to 70°〖KG-*9〗N and SDSF is constructed during a training period of 1954 to 1983. The verification for the hindcast period from 1984 to 2005 suggests that the forecasted DSF is positively correlated with the observation with a maximum correlation coefficient of +0.4; the average Anomaly Correlation Coefficient (ACC) is +0.4 with the root mean square error (RMS) being 1 to 2. Such a result implies that this statistical model has a potential capability in the operational seasonal forecasting of SDSF.