This paper investigates the theory of polarimetric passive remote sensing of wind-generated sea surfaces and the potential application of polarimetric radiometry to ocean wind remote sensing. Theoretical polarimetric emission coefficients of small-scale sea surfaces are evaluated using the small perturbation method (SPM). The SPM is derived to second order and applied to the Stokes vectors of thermal emission from random rough dielectric surfaces described by anisotropic directional spectra. To verify the accuracy of the SPM, a Monte Carlo simulations is performed to calculate the Stokes vectors of the emission from the simulated one-dimensional random rough surfaces with a power law spectrum for various observation angles and surface parameters. The theoretical results of the SPM for all four Stokes parameters are in excellent agreement with the numerical results obtained from the Monte Carlo simulation. Moreover, the second-order coherent fields are indispensable in the theoretical evaluation of the third and fourth Stokes parameters. Otherwise, the reflectivities of random rough surfaces would be significantly overestimated, and the signs of the third and fourth Stokes parameters would be incorrect. The SPM is then applied to small-scale sea surfaces described by an empirical sea surface spectrum. It is found that the azimuthal signatures of Stokes parameters agree qualitatively well with aircraft Ku-band radiometer data. Theoretical model functions of the Stokes parameter are illustrated. Advantages of wind direction retrieval using polarimetric Q and U measurements are discussed, and it is expected that a spaceborne polarimetric radiometer has a potential of providing wind vector measurements with uniform accuracy across all parts of swath. ¿ American Geophysical Union 1994