We present measurements of Ku band electromagnetic (EM) bias made for 2 months from a platform in Bass Strait off the southeast coast of Australia during the austral winter of 1992. EM bias, $epsilon$, the difference between the electromagnetic and true mean sea levels, was measured using Doppler scatterometers. Linear wave theory was used to relate the Doppler signal to the surface displacement, giving simultaneous coincident backscatter and wave measurements, including significant wave height, Hs. On the basis of dimensional reasoning, we suggest that the usual inhomogeneous correlations of the normalized bias ¿ = $epsilon$/Hs with the 10 m wind speed, U10, and Hs may be improved by correlating the data with nondimensional variables, including a characteristic wave slope, s, and wave age, c/U10, where c is a characteristic phase speed of the surface waves. Using both polynomial correlations and optimal estimation techniques to fit the data, we find that the standard error of the fit is reduced by ~50% when the dimensionless variables are used. We find that the dependence of the EM bias on the wave slope is consistent with earlier tower-based measurements and the theory of short-wave modulation by longer waves. We discuss the implications of these results for operational implementation of EM bias algorithms based on wave slope and wave age.