The first space-based Doppler wind lidar (DWL) onboard the Aeolus satellite was launched by the European Space Agency (ESA) on 22 August 2018 to obtain global profiles of horizontal line-of-sight (HLOS) wind speed. In this study, the Raleigh-clear and Mie-cloudy winds for periods of baseline 2B02 (from 1 October to 18 December 2018) and 2B10 (from 28 June to 31 December 2019 and from 20 April to 8 October 2020) were validated using 33 wind profilers (WPRs) installed all over Japan, two ground-based coherent Doppler wind lidars (CDWLs), and 18 GPS-radiosondes (GPS-RSs). In particular, vertical and seasonal analyses were performed and discussed using WPR data. During the baseline 2B02 period, a positive bias was found to be in the ranges of 0.46–1.69 m s⁻¹ for Rayleigh-clear winds and 1.63–2.42 m s⁻¹ for Mie-cloudy winds using the three independent reference instruments. The biases of Rayleigh-clear and Mie-cloudy winds were in the ranges of −0.82−+0.45 m s⁻¹ and −0.71−+0.16 m s⁻¹ during the baseline 2B10 period, respectively. The systematic error for the baseline 2B10 was improved as compared with that for the baseline 2B02. The vertical analysis using WPR data showed that the systematic error was slightly positive in all altitude ranges up to 11 km during the baseline 2B02 period. During the baseline 2B10 period, the systematic errors of Rayleigh-clear and Mie-cloudy winds were improved in all altitude ranges up to 11 km as compared with the baseline 2B02. Immediately after the launch of Aeolus, both Rayleigh-clear and Mie-cloudy biases were small. Within the baseline 2B02, the Rayleigh-clear and Mie-cloudy biases showed a positive trend. For the baseline 2B10, the Rayleigh-clear wind bias was generally negative at all months except August 2020, and Mie-cloudy wind bias gradually fluctuated. The systematic error was close to zero with time in 2020 and did not show a marked seasonal trend. The dependence of the Rayleigh-clear wind bias on the scattering ratio was investigated, showing that the scattering ratio had a minimal effect on the systematic error of the Rayleigh-clear winds during the baseline 2B02 period. On the other hand, during the baseline 2B10 period, there was no significant bias dependence on the scattering ratio. Without the estimated representativeness error associated with the comparisons using WPR observations, the Aeolus random error was determined to be 6.71 (5.12) and 6.42 (4.80) m s⁻¹ for Rayleigh-clear (Mie-cloudy) winds during the baseline 2B02 and 2B10 periods, respectively. The main reason for the large random errors is probably related to the large representativeness error due to the large sampling volume of the WPRs. Using the CDWLs, the Aeolus random error estimates were in the range of 4.49–5.31 (2.93–3.19) and 4.81–5.21 (3.30–3.37) m s⁻¹ for Rayleigh-clear (Mie-cloudy) winds during the baseline 2B02 and 2B10 periods, respectively. By taking the GPS-RS representativeness error into account, the Aeolus random error was determined to be 4.01 (3.24) and 3.02 (2.89) m s⁻¹ for Rayleigh-clear (Mie-cloudy) winds during the baseline 2B02 and 2B10 periods, respectively.