Rhizobium leguminosarum bv. viciae normally gains access to pea host cells through tubular cell wall ingrowths termed infection threads. Matrix glycoprotein (MGP), a major component of the infection thread lumen, is also secreted from the tips of uninoculated roots and can be released into solution under reducing conditions. Monoclonal antibody MAC265, which recognizes MGP through a carbohydrate epitope, was used for immunoaffinity purification of the glycoprotein from pea roots. Following treatment with chymotrypsin, a peptide fragment was obtained and subjected to N-terminal sequencing. Using PCR primers based on this sequence, cDNA clones were isolated with RNA from inoculated roots and nodules. DNA sequencing of 30 of these clones revealed a family of closely related and repetitive polypeptides with (hydroxy)proline-rich motifs. The cDNA sequences showed over 70% identity with the deduced amino acid sequences of plant extensins, particularly with VfNDS-E from Vicia faba and MtN12 from Medicago truncatula, both of which are strongly upregulated in legume root nodules. Root nodule extensins from pea were of variable length but showed strong sequence conservation of the N-terminus, of the C-terminus, and of a central domain comprising 33 amino acids that were sometimes reiterated. The distribution of tyrosine residues suggested the possible importance of intramolecular and intermolecular cross-linking. There was strong sequence conservation with MtN12 in the 3′-untranslated region, suggesting a possible involvement in posttranscriptional regulation of gene expression.