Good progress has been made in forecasting the broad pattern of the weather for periods up to a few days ahead as a result of developments in mathematical weather prediction models. However, our ability to forecast, just a few hours ahead, the detailed weather pattern for specific locations, still leaves a lot to be desired. In this lecture it is suggested that local forecasting, after a quarter century of stagnation, is poised for a decade of slow but steady improvement. Changes will come as a result of new programs which exploit and integrate advances in several areas. These areas include better weather observations, especially from satellites and radar, improved methods of data processing and analysis, and the development of more detailed numerical prediction models. New methods of communications, such as viewdata, will enable the improved local forecasts to be disseminated promptly and in a wide range of specially tailored formats. This will contribute to the development of a more user-oriented meteorological service.