A coder-decoder (codec), based on block coding of transformed coefficients has been proposed for the compression of remotely sensed image data. The technique divided an image into blocks of size nxn and then computed the Discrete W avelet Transform (DWT) of each block at the finest spatial resolution. The blocks of transformed coefficients were classified into types HH, HL, LH, and LL. Each class consisted of group of blocks of size (n/2) (n/2) of similar type. Based on the coefficient variance table of each class, a bit assignment table was computed and blocks of individual class were coded using bit assignment table of the class. The bit rate of compressed data was controllable in this algorithm.