Abstract This study focuses on the ability of ERA-Interim to represent wind variability in the middle atmosphere. The originality of the proposed approach is that wind measurements are deduced from the trajectories of zero-pressure balloons that can reach high-stratospheric altitudes. These balloons are mainly used to carry large scientific payloads. The trajectories of balloons launched above Esrange, Sweden, and Teresina, Brazil, from 2000 to 2011 were used to deduce zonal and meridional wind components (by considering the balloon as a perfect tracer at high altitude). Collected data cover several dynamical conditions associated with the winter and summer polar seasons and west and east phases of the quasi-biennial oscillation at the equator. Systematic comparisons between measurements and ERA-Interim data were performed for the two horizontal wind components, as well as wind speed and wind direction in the [100, 2]-hPa pressure range to deduce biases between the model and balloon measurements as a function of altitude. Results show that whatever the location and the geophysical conditions considered, biases between ERA-Interim and balloon wind measurements increase as a function of altitude. The standard deviation of the model–observation wind differences can attain more than 5 m s−1 at high altitude (pressure P < 20 hPa). A systematic ERA-Interim underestimation of the wind speed is observed and large biases are highlighted, especially for equatorial flights.