The gene encoding full-length human E-cadherin has been cloned and sequenced from liver and colon cDNA libraries (GenBank Accession #L08599). The predicted molecular mass of the unglycosylated and unprocessed protein is 97,000. The human protein conserves most features of the classical cadherins. In its cytoplasmic domain, two approximately 30-35 aminoacid conserved sequence motifs are recognized. These cadherin homology domains have been termed "CH2" and "CH3", and are characteristic of the classical cadherins, but absent or divergent in the more distantly related cadherins such as desmosomal cadherin, T-cadherin, fat, and the human ret oncogene. Given these findings and the importance of cytoplasmic interactions to cadherin function, a subclassification of the cadherin superfamily based on cytoplasmic domain homologies is proposed. This subclassification provides a framework in future studies for understanding the distinct down-stream signaling cascades associated with each cadherin.