Marine winds reported by Voluntary Observing Ships (VOS) and moored buoys require adjustment to provide a homogeneous record of the marine climate. Known sources of inhomogeneity arise from differences in measurement height and method, averaging method and atmospheric stability; methods are available to correct for these. However, significant differences remain in a paired dataset of ship and buoy winds. Regression methods to remove this remaining inconsistency are discussed, and a ranked regression method chosen as most appropriate to adjust ship wind speeds to yield a similar distribution. We show the factors, such as vessel type, that affect the regression results. The corrections, derived from a high-quality paired dataset with rigorous quality control, are effective at reducing inhomogeneity in monthly mean wind speed distributions derived from the International Comprehensive Ocean–Atmosphere Data Set. Copyright © 2005 Environment Canada. Published by John Wiley & Sons, Ltd.