The last 10 years have seen a revolutionary change in the way weather forecasts are made. The basis for today's forecasts is a numerical calculation of the evolution of the energy distribution and motions in a physical model of the atmosphere. The benefits of this change have affected all types of weather forecasting. The substantial benefits to aviation are well established. The benefits to forecasts for the general public can't be conclusively documented but are strongly suggested. Our present limitations are mainly due to insufficient completeness of our atmospheric models. The removal of some of these deficiencies appears to be a relatively straightforward task. Other problems, particularly those of phase change of water, are more difficult. The eventual limitation to the prediction of atmospheric motions appears to be the supply of data describing the initial conditions, but I can safely say that solution of the other problems will require more than just a few years.