Methods for making more efficient use of synthetic aperture radar (SAR) imagery are considered. Local standard deviation and autocorrelation texture measures are used to provide information on the spatial variability in the scattering cross-section. Use of these statistics in a window of 180 × 180m improved classification success rates from 39 to 66 per cent with digitized shuttle imaging radar (SIR-A) data. Muttispectral scanner (MSS) achieves 70 per cent success with the same window size and, by combining this with SAR, a 78 per cent success rate is reached. © 1992 Taylor & Francis Group, LLC.