Near-surface wind over the ocean is a key variable for studies of climate variability and change. Millions of ship reports of near-surface wind are now held in computerized datasets like the Comprehensive Ocean-Atmosphere Data Set (COADS). These data are potentially well suited to many climate studies. However, the reliability of the record is controversial. In particular, the reality of an upward trend in reported wind strength since the 1940s has been questioned and attributed by some to an increasing fraction of anemometer readings on-board ship, compared to reports based on estimation of the sea state. Analyzing the COADS record for 1949–1988, this paper provides evidence on the reliability of the data, corrects the trends in the data, and finally presents some analysis of the multidecadal variability that remains in the data after correction. These multidecadal analyses are offered as a preliminary assessment of the revised wind data and the impact of the corrections. The reliability of the data is assessed by comparing the 10°lat × 10°long seasonal mean reported wind with a wind derived from seasonal mean sea level pressure (SLP) gradients. The wind is derived using the balanced friction flow (BFF) method, which assumes a three-way balance of forces between pressure gradient, Coriolis, and friction, represented using a drag coefficient. The interannual variability of observed and BFF wind generally agree well (except close to the equator, 10°N–10°S), giving good confidence in the data and the BFF theory. However, on the trend timescale, them is considerable disagreement. Globally averaged, the results suggest that there has been an upward trend in reported circulation strength of about 14% over 1949–1988 (highly statistically significant, <1% level), whereas there is no trend of any substance in the globally averaged BFF wind. Since there is an explanation for at least some of the observed wind trend through changes in observational practice and since no serious problems with the BFF trends could be found, the BFF trends have been treated as true and the observed trends corrected to the BFF trends. For this, a smooth field of the trend correction needed at each 10°lat × 10° long box was calculated. Although there is no globally averaged trend in circulation strength in the corrected wind data, there are regional patterns of upward trend (notably in the tropical North Atlantic and the extratropical North Pacific) and downward trend (notably in the equatorial and tropical South Atlantic and the subtropical North Pacific). The changes in the tropical Atlantic are particularly strong in the July–September season, when they accompany the strong downward trend in July–September Sahelian rainfall. The changed wind pattern corresponds to a substantial reduction in cross-equatorial flow. The impact of this on ocean currents and cross-equatorial heat flux should be assessed using ocean numerical models forced with observed wind stress changes. The wind data developed here from a smooth field of corrections are potentially well suited to development for this application. The agreement between BFF winds and observed winds on the interannual timescale was weakest in equatorial latitudes. The utility of the wind data in the equatorial western Pacific is, however, very strongly supported by the close agreement between interannual variability of near-surface divergence, calculated from the corrected near-surface wind, and sea-surface temperature. For interannual studies, the corrections are of minimal importance, so these results, along with the close agreement between BFF and observed winds away from the equator, give a very positive endorsement of the uncorrected COADS's utility for studies of large-scale interannual climate variability.