One application of computer in molecular biology has been the use of its power for comparing a large number of molecular sequences in very short time. For this purpose, a new algorithm is proposed which differs in several aspects from other approaches. Our algorithm, called MAlign, is designed to seek global homology by introducing an effective way to make simultaneous comparisons among test sequences. One problem in previous algorithms which were limited in its ability to compare sequences simultaneously has been solved by introducing intermediate consensus or compacted sequences and including them for comparison. In addition, a homology vector concept was applied to provide uniform representation for each intermediate, which makes global comparison easier. Several test results indicate that high homology values obtained from pairwise alignment are maintained after multiple alignment of those sequences, which is more apparent in higher homology values. Sample alignment results using this approach for three different copper binding proteins as well as bacterial signaling proteins are presented.