Complementary DNA clones encoding the NH2-terminal region of human CR1 have been isolated and sequenced. The deduced complete amino acid sequence of the F allotype of human CR1 contains 2,039 residues, including a 41-residue signal peptide, an extracellular domain of 1,930 residues, a 25-amino acid transmembrane domain, and a 43-amino acid cytoplasmic region. The extracellular domain is composed exclusively of 30 short consensus repeats (SCRs), characteristic of the family of C3/C4-binding proteins. The 28 NH2-terminal SCRs are organized as four long homologous repeats (LHRs) of seven SCRs each. The newly sequenced LHR, LHR-A, is 61% identical to LHR-B in the NH2-terminal two SCRs and greater than 99% identical in the COOH-terminal five SCRs. Eight cDNA clones were spliced to form a single construct, piABCD, that contained the entire CR1 coding sequence downstream of a cytomegalovirus promoter. COS cells transfected with piABCD transiently expressed recombinant CR1 that comigrated with the F allotype of erythrocyte CR1 on SDS-PAGE and that mediated rosette formation with sheep erythrocytes bearing C4b and C3b. Recombinant CR1 also had factor I-cofactor activity for cleavage of C3(ma). Analyses of six deletion mutants expressed in COS cells indicated that the NH2-terminal two SCRs of LHR- A contained a site determining C4 specificity and the NH2-terminal two SCRs of LHR-B and -C each had a site determining C3 specificity. The presence of these three distinct sites in CR1 may enable the receptor to interact multivalently with C4b/C3b and C3b/C3b complexes generated during activation of the classical and alternative pathways.