Post-translational isomerization of l-amino acids to d-amino acids is a subtle modification, not detectable by standard techniques such as Edman sequencing or MS. Accurate predictions require more sequences of modified polypeptides. A 46-amino-acid-long conotoxin, r11a, belonging to the I-superfamily was previously shown to have a d-Phe residue at position 44. In this report, we characterize two related peptides, r11b and r11c, with d-Phe and d-Leu, respectively, at the homologous position. Electrophysiological tests show that all three peptides induce repetitive activity in frog motor nerve, and epimerization of the single amino acid at the third position from the C-terminus attenuates the potency of r11a and r11b, but not that of r11c. Furthermore, r11c (but neither r11a nor r11b) also acts on skeletal muscle. We identified more cDNA clones encoding conopeptide precursors with Cys patterns similar to r11a/b/c. Although the predicted mature toxins have the same cysteine patterns, they belong to two different gene superfamilies. A potential correlation between the identity of the gene superfamily to which the I-conotoxin belongs and the presence or absence of a d-amino acid in the primary sequence is discussed. The great diversity of I-conopeptide sequences provides a rare opportunity for defining parameters that may be important for this most stealthy of all post-translational modifications. Our results indicate that neither the chemical nature of the side chain nor the precise vicinal sequence around the modified residue seem to be critical, but there may be favored loci for isomerization to a d-amino acid.