The primary difference between MSA and previous multiple alignment programs is its capability to align more than three sequences using an explicitly defined measure of overall alignment quality. The default measure used by MSA considers some replacements more costly than others and penalizes for gaps in a manner reflecting the fact that a single mutational event can insert or delete several residues (14, 25). With pairwise alignment costs so defined, MSA computes a multiple alignment that minimizes the sum of the pairwise costs, weighting the pairs using information derived from an evolutionary tree. This approach is most effective in aligning sequences that share a global, but perhaps quite distant, relationship. Other methods may be more appropriate for the analysis of the statistical significance of sequence similarities (8), detection of sequence motifs or consensus sequences (3, 9, 22), or aligning large numbers of sequences (17-21). Although MSA will not always produce alignments in such good agreement with structural super positions as seen here, we believe it can be a powerful sequence analysis tool for molecular biologists.