Relaxin family peptides perform a variety of biological functions by activating four G protein-coupled receptors, namely RXFP1-4. Among these receptors, RXFP3 lacks a specific natural or synthetic agonist at present. A previously designed chimeric R3/I5 peptide, consisting of the B-chain of relaxin-3 and the A-chain of INSL5, displays equal activity towards the homologous RXFP3 and RXFP4. To increase its selectivity towards RXFP3, in the present study we conducted extensive mutagenesis around the B-chain C-terminal region of R3/I5. Decreasing or increasing the peptide length around the B23-B25 position dramatically lowered the activation potency of R3/I5 towards both RXFP3 and RXFP4. Substitution of B23Gly with Ala or Ser converted R3/I5 from an efficient agonist to a strong antagonist for RXFP3, but the mutants retained considerable activation potency towards RXFP4. Substitution of B24Gly increased the selectivity of R3/I5 towards RXFP3 over the homologous RXFP4. The best mutant, [G(B24)S]R3/I5, displayed 20-fold higher activation potency towards RXFP3 than towards RXFP4, meanwhile retained full activation potency at RXFP3. Thus, [G(B24)S]R3/I5 is the best RXFP3-selective agonist known to date. It is a valuable tool for investigating the physiological functions of RXFP3, and also a suitable template for developing RXFP3-specific agonists in future.