Wind velocity data from seven high-altitude long-duration balloon flights are compared with stratospheric analyses and forecasts produced by the U.K. Meteorological Office (UKMO) and the Data Assimilation Office (DAO). The results suggest that biases in both the UKMO and DAO analyses arise from the displacement of the polar night jet to higher latitudes in the winter and from a cold bias in the stratosphere in the summer. At high latitudes the DAO analysis performs better than the UKMO analyses, with the use of a rotated pole in the DAO system the most likely cause of the improvement. The trajectories for each flight were simulated using UKMO and DAO 5-day forecasts and by using simple persistence techniques. It is shown that UKMO and DAO forecasts based on better initial analyses produced trajectories that were more accurate. On the 1-day forecast timescale there is little difference between the techniques, but on the 2- to 5-day range the trajectories based on forecasts generally performed significantly better than persistence. The results demonstrate the usefulness of stratospheric balloon data for validating both analyses and forecasts. ¿ 2001 American Geophysical Union