The complete sequence of 595 amino acids of the alpha-I domain of human erythrocyte spectrin has been determined. Peptides derived from three different protease cleavages were purified using high performance liquid chromatography and subjected to automated amino acid sequence analysis. These data along with sequences of the cyanogen bromide and large tryptic peptides (Speicher, D.W., Davis, G., Yurchenco, P.D., and Marchesi, V.T. (1983) J. Biol. Chem. 258, 14931-14937) represent most or all of the sequence of spectrin alpha-I. The single remaining ambiguity is the precise termination of the COOH terminus of the alpha-I domain. The sequence data suggest that the 595 residues presented here represent the complete sequence of the alpha-I domain, but the apparent size of the COOH-terminal CNBr fragment suggests the existence of an additional 38 residues at the end of the domain. The sequence of the alpha-I domain contains a single type of internal homology composed of multiple 106-amino acid repeats consistent with the occurrence of multiple gene duplications during the course of spectrin evolution. The only portion of the alpha-I sequence which does not appear to contain this sequence repeat is the segment containing the NH2-terminal 17 residues. This unique segment may be part of the oligomer binding site. No disulfide bonds appear to be involved in the structure of alpha-I and cysteine is not highly conserved. Calculations of secondary structure suggest the presence of short helices which fold into triple helical segments approximately 50 A in length. There is little beta sheet structure. A model of spectrin structure incorporating the repeat unit and proposed secondary structure is presented. A computer search of alpha-I sequence with the National Biomedical Research Foundation database of 2145 protein sequences did not detect any significant relationships. Spectrin is apparently the first member of a new class of proteins to be structurally characterized.