A method to diagnose the dominant forcing direction in locally coupled ocean-atmosphere anomalies is applied to the Global Reanalysis, to an NCEP-AMIP run and to the MRF reforecast data to examine the implications of using ocean-driving scenarios for atmospheric simulation and prediction. The method applied to the Reanalysis confirms that the atmosphere tends to drive the ocean in the extratropics, whereas the reverse is true in the tropics. It is also found, for the first time, that the longer-lasting extratropical and tropical anomalies tend to have, respectively, atmosphere-driving and ocean-driving phase relationships. As a consequence, and since the AMIP run tends to ignore the atmosphere-driving cases, simulations with prescribed SST tend to shorten the duration of atmospheric anomalies in the extratropics and lengthen their duration in the tropics, as compared to the Reanalysis. Both AMIP run and the MRF tend to have a lower skill of simulation and prediction in cases when the method diagnoses a predominant atmosphere-driving situation in the Northern Pacific. Potential use of this method to analyze coupled land-atmosphere anomalies is discussed.