Abstract. Ocean surface, grazing-angle radar backscatter data from two separate experiments, one of which provided coincident time series of measured surface winds, were found to exhibit signatures of deterministic chaos. Evidence is presented that the lowest dimensional underlying dynamical system responsible for the radar backscatter chaos is that which governs the surface wind turbulence. Block-averaging time was found to be an important parameter for determining the degree of determinism in the data as measured by the correlation dimension, and by the performance of an artificial neural network in retrieving wind and stress from the radar returns, and in radar detection of an ocean internal wave. The correlation dimensions are lowered and the performance of the deterministic retrieval and detection algorithms are improved by averaging out the higher dimensional surface wave variability in the radar returns.