The homologous repeats of annexin tetrads are believed to have originated by successive duplication and fusion from a putative monomeric precursor, but neither the nature of their ancestor nor the events leading to the formation of different subfamilies have been elucidated. We have performed molecular phylogenetic analysis of aligned annexin nucleotide and amino acids sequences to characterize subfamily branching, to delineate the temporal order of appearance of individual repeat units, and to gain insight into the origin and nature of the primordial unit. All extant annexins appear to have a common tetrad precursor that may have originated from a progenitor unit resembling repeat 3, followed by the generation of repeats 4, 1, and 2 from a more evolved progenitor with subsequent fusion. Repeat sequences of the earliest human annexins VII and XIII were used to identify alpha-giardin genes as primitive homologues from the unicellular protozoan Giardia lamblia, which diverged from eukaryote lineage 1-1.5 billion yr ago. The significant homology between alpha-giardins and annexins suggested that the cell membrane adhesive role of these proteins may be a common, fundamental property of the annexin C-terminal core region. Purported annexin VII of Dictyostelium discoideum was reclassified as new annexin XIV, three Caenorhabditis elegans genes were assigned to new subfamilies XV, XVI, and XVII, and plant annexin XVIII from Medicago sativa was among the earliest diverging subfamilies. Annexins I and II were found to be closely related, but analysis of protein mutation rates confirmed that the former is evolving up to three times more rapidly. The inclusion of early phyla in annexin taxonomy provides a useful basis for assessing the structural and functional changes associated with annexin evolution.