Research applicable to measuring winds with a scatterometer and the subsequent use of the data is briefly summarized. Data that demonstrate that (U(λ/2)/C(λ))-1, which is a quantity closely related to the normal stress, is more closely correlated to the nondimensionalized wave spectrum at L band, (ϕ(ω)ω5/g2), than either U(19.5) or u*, which is related to the shear stress, are given. The SOS wind recovery algorithm is shown to fail for light winds (given the power law assumption) and to bias climatological wind data. The data from a scattterometer are far too dense to be used in a synoptic scale analysis. Properties of superobservations to be used for synoptic scale analysis at a 1¿ resolution are derived and shown to produce very accurate wind fields, given that the scatterometer winds are correct. Superobservations are shown to yield reliable estimates of the horizontal component of the divergence and the curl of the wind stress. Disadvantages of a plus or minus 3-hour data window for scatterometer data assimilation are describe