The amino acid sequences of 22 alpha-amylases from family 13 of glycosyl hydrolases were analyzed with the aim of revealing the evolutionary relationships between the archaeal alpha-amylases and their eubacterial and eukaryotic counterparts. Two evolutionary distance trees were constructed: (i) the first one based on the alignment of extracted best-conserved sequence regions (58 residues) comprising beta 2, beta 3, beta 4, beta 5, beta 7, and beta 8 strand segments of the catalytic (alpha/beta)(8)-barrel and a short conserved stretch in domain B protruding out of the barrel in the beta 3 --> alpha 3 loop, and (ii) the second one based on the alignment of the substantial continuous part of the (alpha/beta)(8)-barrel involving the entire domain B (consensus length: 386 residues). With regard to archaeal a-amy lases, both trees compared brought, in fact, the same results; i.e., all family 13 alpha-amylases from domain Archaea were clustered with barley pi isozymes, which represent all plant alpha-amylases. The enzymes from Bacillus licheniformis and Escherichia coli, representing liquefying and cytoplasmic alpha-amylases, respectively, seem to be the further closest relatives to archaeal alpha-amylases. This evolutionary relatedness clearly reflects the discussed similarities in the amino acid sequences of these alpha-amylases, especially in the best-conserved sequence regions. Since the results for alpha-amylases belonging to all three domains (Eucarya, Eubacteria, Archaea) offered by both evolutionary trees are very similar, it is proposed that the investigated con-served sequence regions may indeed constitute the "sequence fingerprints" of a given alpha-amylase.