A data compression and decompression procedure for a serial printer handles data from a print swath (30) that includes Y rows of pixel data. The procedure divides the print swath (30) into N microbands (32, 34, 36...) of pixel data, with each microband (30) including y rows of pixel data (where y x N ≥ Y). The procedure compresses each microband (30) of the pixel data by breaking up the microband (30) into microblocks (38) and comparing data in each microblock (38) with a seed value to arrive at a compressed microblock (54) based upon data differences from the seed value. Each microblock (38) comprises a block of pixel data that is m bits wide by y rows high, where m is a small fraction of M bits of pixel data that are printable by the printhead across a sheet. All microbands (32, 34, 36...) of a swath (30) are subjected to the data compression procedure so as to create positionally corresponding sets of compressed data microblocks (54) for all of the microbands (32, 34, 36...). In preparation for printing, the procedure decompresses all N microbands (32, 34, 36...) of raster data by acting on each set of compressed microblocks (54) to recreate a set of N microbands (32, 34, 36...) of decompressed pixel data. The procedure then feeds the N microbands (32, 34, 36...) of decompressed raster data, in parallel, to the printhead (10, 12, 14, 16).