Improved methods of forecasting wind strengths between two and 24 hours ahead have long been a demand of utilities to make wind energy a more attractive prospect. Such forecasting helps utilities with scheduling of thermal plant, but predicting weather is an inexact science. Two projects in the Joule III programme aim to improve it. A project lead by Denmark's Risø National Laboratory has drawn together prediction models from a number of sources and a single technique, combining the advantages of each, is now being tested. A similar collaboration between the French School of Mines and the Technical University of Crete is also tackling the problem. It is testing the accuracy of the predictions against data from the Greek Islands of Lemnos and Crete. The Risø study also aims to quantify the economic benefits associated with improved predictability methods.