Abstract. Comparisons are made between horizontal wind measurements carried out using a VHF-radar system at Aberystwyth (52.4°N, 4.1°W) and radiosondes launched from Aberporth, some 50 km to the southwest. The radar wind results are derived from Doppler wind measurements at zenith angles of 6°in two orthogonal planes and in the vertical direction. Measurements on a total of 398 days over a 2-year period are considered, but the major part of the study involves a statistical analysis of data collected during 75 radiosonde¯ights selected to minimise the spatial separation of the two sets of measurements. Whereas good agreement is found between the two sets of wind direction, radar-derived wind speeds show underestimates of 4±6% compared with radiosonde values over the height range 4±14 km. Studies of the characteristics of this discrepancy in wind speeds have concentrated on its directional dependence, the eects of the spatial separation of the two sets of measurements, and the in¯uence of any uncertainty in the radar measurements of vertical velocities. The aspect sensitivity of radar echoes has previously been suggested as a cause of underestimates of wind speeds by VHF radar. The present statistical treatment and case-studies show that an appropriate correction can be applied using estimates of the eective radar beam angle derived from a comparison of echo powers at zenith angles of 4.2°a nd 8.5°.