Images are now widely available in the form of digital data. Unfortunately the amount of data that results from the digitization process is extremely large, resulting in high costs in terms of storage demand or transmission bandwidth for such data. Image data compression techniques are becoming widely used in order to reduce these problems, a standard technique being the Discrete Cosine Transform (DCT). Significant compression ratios are achievable using the DCT (typically 20:1), but with the drawback of some loss of accuracy in the decompressed reconstructed image. In some applications (e.g., videophone) some loss in the received data can be tolerated. The purpose of this paper is to present an assessment of the compression available when the DCT is applied to remote sensing images, and to determine the effects of data loss in terms of image fidelity and experimental multi-band classification of reconstructed image data. Results are presented to demonstrate the effects of using the DCT for remote sensing image data compression.