Abstract: Based on the hindcast data of the second generation monthly dynamic extended range forecast model (DERF2.0) of National Climate Center and historical spring sowing data, combined with NCEP/NCAR reanalysis data, correlation analysis is conducted between historical spring sowing data and reanalysis data and model hindcast data, respectively. Regions with anomalies correlation coefficient (ACC) passing significant test are defined as key circulation zones, and the overlapping areas where both anomalies correlation coefficients passing the significant test are selected as key influencing factors of climatic conditions of spring sowing period in the middle and lower reaches of the Yangtze. Using historical time series of selected factors of certain circulation variables, e.g., geopotential height of 200, 500 hPa and 700 hPa, zonal and meridional wind of 850 hPa as predictors, favorable, unfavorable and continuous unfavorable days of spring sowing period in the middle and lower reaches of the Yangtze as predictors, utilizing optimal subset regression method (OSR), a model interpretation scheme of climatic conditions of sowing period prediction is established. The performance of the model interpretation scheme with different lead time are evaluated and analyzed. Meanwhile, the predictive skill of typical years with unfavorable climatic conditions is tested. Hindcast test of predictive scheme exhibits considerable overall predictive skill on both favorable and unfavorable days of spring sowing period. Predictive results of different lead time indicate that the prediction performance of unfavorable and continuous unfavorable days grows better as the lead time shortens. Moreover, hindcast result with lead time equals 0 shows that the model interpretation scheme not only simulates the annual variations well, but also illustrates certain predictive ability of decadal change of climatic conditions of spring sowing period. In the operational prediction of climatic conditions of spring sowing period, rolling forecast result of model interpretation scheme should be utilized to achieve better predictive performance. Since the model interpretation scheme does not give climatic conditions of spring sowing period directly, each variables of model output should be considered. In order to test the predictive skill of typical years with unfavorable climatic conditions, five years with typical unfavorable climatic conditions of spring sowing period (with continuous unfavorable days exceeding 10 days) are selected and verification results indicate that since the 1980s, the integrated result of scheme is approximately the same with observation, which exhibits considerable predictive skill.