Previous studies have revealed the propensity of elastin-based biopolymers to form amyloid-like fibers when dissolved in water. These are of interest when considered as "ancestral units" of elastin in which they represent the simplest sequences in the hydrophobic regions of the general type XxxGlyGlyZzzGly (Xxx, Zzz = Val, Leu). We normally refer to these biopolymers based on elastin or related to elastin units as "elastin-like polypeptides". The requirement of water for the formation of amyloids seems quite interesting and deserves investigation, the water representing the natural transport medium in human cells. As a matter of fact, the "natural" supramolecular organization of elastin is in the form of beaded-string-like filaments and not in the form of amyloids whose "in vivo" deposition is associated with some important human diseases. Our work is directed, therefore, to understanding the mechanism by which such hydrophobic sequences form amyloids and any conditions by which they might regress to a non-amyloid filament. The elastin-like sequence here under investigation is the ValGlyGlyValGly pentapeptide that has been previously analyzed both in its monomer and polymer form. In particular, we have focused our investigation on the apparent stability of amyloids formed from poly(ValGlyGlyValGly), and we have observed these fibers evolving to a hydrogel after prolonged aging in water. We will show how atomic force microscopy can be combined with X-ray photoelectron spectroscopy to gain an insight into the spontaneous organization of an elastin-like polypeptide driven by interfacial interactions. The results are discussed also in light of fractal-like assembly and their implications from a biomedical point of view.