During EASOE 10 day backward trajectories ending on most measurement sites were calculated on 6 isentropic levels every day. The temporal and spatial coarseness of the ECMWF analyses used cause significant errors in the trajectories. These interpolation errors are largest in the direction along the trajectories. The difference between two trajectories of the Polar Vortex Balloon Experiment and calculated trajectories is 8 and 12% of the trajectory length after 1.83 and 2.14 days. This discrepancy is not explainable for interpolation errors alone. ¿ American Geophysical Union 1994