Accurate wind forecasts will be essential in the utilization of wind energy conversion systems (WECS). In order to assess the reliability at present forecast methods for wind speed forecast data from the USA and Sweden have been verified. Data represent different methods, numerical/statistical and subjective, different locations, seasons and heights. However, the data sets are too small to allow any definite conclusions. The results point to that none of the tested forecast methods meet the requirements on forecast error put forward by utilities. The best forecasts were obtained by subjective methods based on numerical prediction for projection times less than +18h. Beyond that time objective, numerical/statistical methods showed to be better. National weather services are recommended to improve forecast methods for shortrange forecasts, 0-12 hours ahead. To obtain sufficient forecast accuracy future WECS sites must also supply relevant observations of low level atmospheric structure.