Recent advancements in radar sensor technology and processing capabilities enable a large volume of data to be collected and made available to the war fighter. However, the bandwidth of data dissemination and storage is strained. Synthetic aperture radar (SAR) imagery, different from other remote sensors, is complex, containing not only magnitude but also phase components, i.e., two channels per pixel. This paper describes Lockheed Martin's wavelet/trellis coded quantization (W/TCQ) compression algorithm and presents results of quantitative assessment of the impact of data compression on complex SAR data.Compression performance was evaluated both quantitatively using statistical measurements and qualitatively by image analysts. Trade studies of various compression schemes and data formats were also investigated. Data distortion due to compression/decompression was assessed using statistical metrics such as signal-to-quantization noise ratio, RMS phase distortion, and ability to support subsequent complex data processing operations. Quantitative assessments were performed with error analyses and a simple decision function to address how data compression affects ATR performance.Data compression did not significantly reduce image utility for magnitude channel compression ratios of up to 64:1. On the other hand, the phase channel compression required 4 bits per pixel to allow adequate phase fidelity to support subsequent complex operations. Magnitude/phase representation afforded better compression results than representing the data in I/Q format. Should this finding remain consistent, it has a significant impact on the compression scheme employed in an operational scenario. The magnitude and phase channels can be allocated different compression ratios to maintain both high quality image interpretability and necessary phase fidelity. Index terms - Compression, complex SAR, trellis coded quantization, radar