The European Space Agency's ADM-Aeolus mission will for the first time provide wind profile measurements from space. In this study we carry out global simulations to predict the yield (i.e. quantity) and accuracy of future ADM-Aeolus wind profile measurements, using realistic clouds and climatological aerosol distributions. Based on an assessment of wind analysis accuracy, we argue that the main Aeolus impacts on global data assimilation are to be expected (i) in the jet streams over the oceans, especially away from main air traffic routes, and in the African/Asian subtropical jet, (ii) in the lower troposphere, e.g. western parts of the N. Pacific and N. Atlantic oceans, if cloud gaps are sufficient, and (iii) in the tropics, where mass–wind balance is weak and hence temperature information is not effective for inferring wind. These expectations are supported by the simulations of the yield and accuracy of ADM-Aeolus data. Observational data from the 1994 Lidar In-space Technology Experiment (LITE) are used to provide realistic profiles of cloud cover as input to such simulations. Profiles of model cloud cover are also used and these are validated against the LITE data. While the overall occurrence of model cloud agrees well with LITE-inferred cloud, model cloud cover underestimates the observations by around 20%. The use of model cloud cover for simulating Aeolus data thus results in a modest overestimation of Aeolus penetration altitude. Probability distributions of Aeolus instrument error show good agreement between simulations with observed and model cloud, but the worst 10% of errors are underestimated when model cloud cover is used. The mission-specified accuracy requirement for the free troposphere (instrument error below 2 m s−1) is met by more than 90% of the simulated data. The corresponding requirement for the boundary layer (instrument error below 1 m s−1) is met by two thirds of simulated data from the Mie channel, even when substantial cloud is encountered and under conservative assumptions about aerosol backscatter. It is shown that, after accounting for errors of representativeness, such data can be expected to receive appreciable weight in global data-assimilation systems. Copyright © 2005 Royal Meteorological Society