High-precision runoff prediction is crucial for reducing water-related hazard risks. Sub-seasonal to seasonal (S2S) prediction can provide more helpful information due to its longer lead time. As the upper-stream boundary of the basin, the headwater of Yangtze River plays an important role in the runoff prediction of the whole basin. S2S prediction in this area can further extend the timeliness of prediction. However, the existing S2S meteorological predictions often have large errors in this area, and have coarse spatial resolutions, which may cause large biases when directly used as the forcing data of hydrological models with high requirements for data quality. Therefore, how to improve the effect and accuracy of S2S meteorological and runoff prediction in the headwater of Yangtze River is worthy of attention. In this study, the bias-correction effect on S2S prediction and the impacts of two correction strategies (pre- and post-processing) on S2S runoff prediction for the headwater of Yangtze River are analyzed. S2S precipitation and temperature products from four centers are evaluated in the Tibetan Plateau. Then, the S2S products are bias corrected and the changes in prediction effects after correction are analyzed. The Xin’an River model established in the headwater of Yangtze River is driven by the S2S products, and the impacts of pre- and post-processing on runoff prediction are compared. The results show that S2S product from ECMWF performs the best for precipitation prediction. When predicting temperature, S2S product from UKMO performs best. S2S product from ECMWF performs well in predicting minimum temperature but has large error in predicting maximum temperature. Bias correction can improve the ability of S2S products from all centers to predict precipitation and temperature significantly. The correlation coefficients between sub-seasonal runoff predictions and observation improve and the error decreases after bias correction. Meanwhile, the correction effect is more obvious with the growth of lead time. Taking both the correction effect and calculation steps into account, correcting runoff is more suitable for the sub-seasonal runoff prediction correction of the headwater of Yangtze River.