Progress continues to be made on an automated method to estimate the intensity of North Atlantic tropical cyclones (TC) using Special Sensor Microwave Imager (SSM/I) digital data. Until recently, this effort focused exclusively on the features extracted from the 85 GHz channel ice scattering brightness temperatures (Tb). These features are used to represent the current set of 60 TC's (319 image samples) in a machine learning algorithm. Feature selection followed by leave-one-out cross validation resulted in an RMSE of 15.0 kts. Applying a gradient vector axisymmetry method (adapted from research at the University of Arizona) to the data as well as extracting features from the 37 GHz channel (associated with lower level cloud liquid water) improved the RMSE to 11.9 kts. In the gradient vector axisymmetry method, the deviation of the gradient vector angle (calculated at every pixel in a two-degree radius) from a radial extending from the TC center is determined. The variance of the deviation angles is calculated as the axisymmetry measure. Stronger storms have more organized rainbands and eyewall features and produce lower variance values (implying higher axisymmetry). Weaker storms have disorganized rainbands resulting in variance values that are higher (lower axisymmetry). This feature was computed for both 85 GHz and 37 GHz data. Other 37 GHz “features” include general measurements (e.g., average Tb in inner 1-degree radius) as well as more TC structure-related and image organization features, including percent encirclement of higher Tbs above certain thresholds (determined through histogram analysis), Tb difference of eye and the surrounding environment, simple symmetry measures, etc. Future work will involve adding samples to the data set from different years and different passive microwave sensors and determining criteria that will enable these intensity estimates to be included in the Satellite Consensus (SATCON) method developed by the Cooperative Institute for Meteorological Satellite Studies (CIMSS).