SH3 domains bind proline-rich peptides with affinities in the order of 0.2-50 microM. In general, these domains are quite promiscuous, and the same peptide can bind to several different SH3 domains with similar affinities (i.e., 3BP1 peptide to Abl- and Fyn-SH3). This poor affinity and specificity make it difficult to elucidate their role in vivo as well as the use of peptides to specifically bind to a single domain. Here, we report that by using existing biocomputing tools, as well as simple physicochemical reasoning, it is possible to design mutations in the 3BP1 peptide (Met4-Tyr, Pro5-Ser, and Leu8-Pro), so that the affinity for Abl-SH3 increases 20-fold (p40 peptide: APTYSPPPPP; Kd = 0.4 microM), while that for the closely related domain, Fyn-SH3, decreases 10-fold. Both the RT and n-Src loops are responsible for regulating the specificity for Pro-rich ligands and more specifically residues Ser15, Thr19, and Glu38 in Abl-SH3. The first six positions in the 3BP1 peptide are important for determining the specificity for SH3 domains, while the remaining four seem to be more important for the affinity. Moreover, by choosing rationally the substituents, it is possible to replace some of the Pro residues postulated to be essential for the interaction with SH3 domains and still have a significant affinity. This indicates that the sequence repertoire that could interact with a specific SH3 domain could be larger than previously thought.