The effect of the vector quantization (VQ) codebook (and codevector) size on the quality of synthetic-aperture radar (SAR) images formed from the compressed phase history data are shown. The reconstructed SAR images are compared to the original SAR image using the terrain and contrast ratios as measures of quality. Experiments indicate that image quality improves monotonically with increasing codevector size. The authors infer that compression ratios of 10:1 may be feasible for many applications.<>