A statistical postprocessing algorithm (i.e., the decaying average method) has been applied to the NCEP/GEFS and CMC/GEFS to generate calibrated forecasts. The implementation of this technique is expected to improve NCEP and CMC global ensemble forecasts in order to provide more accurate NAEFS products. Due to the different ensemble configurations, calibration strategies applied to the NCEP and CMC ensembles have been adjusted. The NCEP/GEFS is created by using one model with perturbed initial conditions. We assume the biases from one model have a kind of similarity, and ensemble mean biases are thought to be able to represent these systematic errors. Therefore, NCEP/GEFS uses the ensemble mean bias to calibrate each member. For CMC/GEFS, the bias for each individual ensemble member is calculated and used for that member. Both the NCEP/ GEFS and CMC/GEFS benefit from the application of bias correction. Several studies have shown that NAEFS, when compared to the CMC and NCEP ensemble system, shows significant improvements both in terms of reliability and resolution (Zhu and Toth 2008; Candille 2009). Even with the attractive properties of the decaying average method, its limitations and performance for some variables in week 2 forecasts in some seasons represent a major drawback. There is room for future improvement from (a) adjusting weights to allow a longer training time and (b) to take advantage of reforecasts/hindcasts.