Most operational extended range forecasting models assume that the ocean always forces the atmosphere. To a certain degree, this approach results in a beneficial impact on intraseasonal and interannual predictability. However, the experience with Atmospheric Model Intercomparison Project (AMIP) runs, which also use the one-way interaction approach, suggest that the feedback effect of atmosphere to the ocean is not negligible. Moreover, observational studies indicate that the atmosphere on average drives the ocean over the extratropics. Using the NCEP/NCAR Reanalysis we have generated a climatology of the types of interactions between the ocean and the atmosphere over the globe. The climatology contains the statistics of the frequency of ocean-driving versus atmosphere-driving anomalies in the ocean-atmosphere system with time scales ranging from a week to a season. For each anomaly the forcing direction, either from the ocean or from the atmosphere, was determined using a dynamical rule. The results are further verified with an independent method that examines the temporal phase relation between the atmospheric and oceanic anomalies and with the traditional cross correlation between the anomalies. A similar procedure was carried out using an NCEP-AMIP run data. Differences between the two climatologies are found over the extratropics with a general tendency of the AMIP data to have a fewer number of atmosphere-driving anomalies than the Reanalysis.