Ice accumulation on overhead railway wires can damage railway devices and disrupt railway services. In order to provide timely short-period forecasts of ice on the wires, an automated wire ice-prediction model is developed in this study. It is a statistical model with much simplified mechanisms of icing, and provides forecasts of wire surface temperature and wire surface state (icy or not), along with the forecasts of air temperature, dew point and wind speed, for three hours ahead. The model was validated against actual sensor measurements at the Dalton test site in north-east England during the winter of 1998-99. The results of the validation show that the model forecasts of wire surface temperature, air temperature, dew point and wind speed are slightly biased. The standard deviation of wire surface temperature forecast is 1.43°C. The accuracy of ice prediction when ice was reported is 81.8% and the accuracy of no-ice forecast when ice was not observed is 97.5%.