We present a new methodology to identify SST fronts of the Gulf Stream, using linear relationships between sea surface temperature (SST) gradients, and curl and divergence of wind stress fields, derived from high resolution (1 km) SAR data. The new approach uses a composite metric determined from the wind stress divergence and curl fields from individual SAR images. Multi-stage spatial filtering, Wiener and Gaussian low-pass filters, and a statistically-based high pass spatial filter are applied to the derived wind stress curl and divergence fields. Results are significantly improved by restricting SAR imagery to cases where wind speed is less than 12 m/s, thus removing strong wind shear fronts. The method is demonstrated with SAR images of the Gulf Stream and has potential to be applied in near real time operations. The advantages of SAR imagery over optical sensors are its independence of cloud or night-time conditions and high accuracy.