This paper describes the application of dynamic combinatorial libraries (DCL) towards the discovery of self-assembling nanostructures based on aromatic peptide derivatives and the continuous enzymatic exchange of amino acid sequences. Ultimately, the most thermodynamically stable self-assembling structures will dominate the system. In this respect, a library of precursor components, based on N-fluorenyl-9-methoxycarbonyl (Fmoc)–amino acids (serine, S and threonine, T) and nucleophiles (leucine, L–; phenylalanine, F–; tyrosine, Y–; valine, V–; glycine, G–; alanine, A–OMe amino-acid esters) were investigated to produce Fmoc–dipeptide esters, denoted Fmoc–XY–OMe. Upon exposure to a protease (thermolysin), which catalyses peptide bond formation and hydrolysis under aqueous conditions at pH 8, dynamic libraries of self-assembling gelator species were generated. Depending on the molecular composition of the precursors present in the library different behaviours were observed. Single components, Fmoc–SF–OMe and Fmoc–TF–OMe, dominated over time in Fmoc–S/(L+F+Y+V+G+A)–OMe and Fmoc–T/(L+F+Y+V+G+A)–OMe libraries. This represented >80% of all peptide formed suggesting that a single component molecular structure dominates in these systems. In a competition experiment between Fmoc–(S+T)/F–OMe, conversions to each peptide corresponded directly with ratios of starting materials, implying that a bi-component nanostructure, where Fmoc–TF–OMe and Fmoc–SF–OMe are incorporated equally favourably, was formed. Several techniques including HPLC, LCMS and fluorescence spectroscopy were used to characterize library composition and molecular interactions within the self-selecting libraries. Fluorescence spectroscopy analysis suggests that the most stable peptide nanostructures show significant π–π intermolecular electronic communication. Overall, the paper demonstrates a novel evolution-based approach with self-selection and amplification of supramolecular peptide nanostructures from a complex mixture of amino acid precursors.