The wind speeds obtained from the Seasat A scatterometer system (SASS) and scanning multichannel microwave radiometer (SMMR) using two different algorithms were compared on a global scale. The temperature dependence of the sea surface emissivity was shown to be incorrectly modelled. After correcting this effect, regional differences up to ¿3 m s-1 are still observed between both instruments, even though they balance in global averaging, resulting in no bias between the global data sets. Validation experiments of satellite wind speeds should take into account this possibility of regional biases and insure the validity of the measurements everywhere in the global ocean. ¿ American Geophysical Union 1990