This paper suggests a new test data compression scheme that performs Huffman coding on different sections of test data file separately. It improves upon the single Huffman tree based approach by upto 6% in compression ratio, 29% in test application time, sacrificing only 6.1%in the decoder area. The scheme compares favourably with other works reported in the literature. While for most of the cases, it produces better compression ratios, the area requirements are much lesser than other contemporary works.