A seismic survey can produce a vast amount of data which is on the order of hundreds of terabytes. A high performance seismic data decompression algorithm can solve problems related not only to storage and transmission but also bandwidth. Decoding is one of the most expensive computational process in seismic data decompression due to the use of enthropy encoding algorithms as Huffman or arithmetic coding. We have designed a Huffman decoder and implemented it on an FPGA. Results indicate that the operating frequency, throughput and use of logical resources in our implementation are better than other predecessors. The design exploits the advantages of the FPGAs to process data at the bit level and so that it can operate at frequencies up to 265 MHz and uses less than 1% of the FPGA logic resources.