Abstract: Depending on the scale of observation, many engineered and natural materials show different mechanical behaviour. Thus, size effect theories, based on a multiscale approach, analyse the intrinsic (due to microstructural constraints, e.g., grain size) and extrinsic effects (caused by dimensional constraints), in order to improve the knowledge in materials science and applied mechanics. Nevertheless, several problems regarding Solid Mechanics and Materials Science cannot be solved by conventional approaches, because of the complexity and uncertainty of materials proprieties, especially at different scales. For this reason, a simple model, capable of predicting a fracture toughness at different scale, has been developed and presented in this paper. This model is based on the Golden Ratio, which was firstly defined by Euclide as: “A straight line is said to have been cut in extreme and mean ratio when, as the whole line is to the greater segment, so is the greater to the less”. Intimately interconnected with the Fibonacci sequence (1, 2, 3, 5, 8, 13, …), this number controls growth in Nature and recurs in many disciplines, such as art, architecture, design, medicine, etc.., and for man-made and natural brittle materials, the Golden Ratio permits to define the relationship between the average crack spacing and the thickness of quasi-brittle materials. In these cases, the theoretical results provided by the Golden Ratio, used to calibrate a size-effect law of fracture toughness, are in accordance with the experimental measurements taken in several test campaigns carried on different materials (i.e., rocks, ice, and concrete). This paper presents the case of fracture toughness of snow, in which the irrational number 1.61803 recurs when the geometrical dimensions vary. This aspect is confirmed by the results of experimental campaigns performed on snow samples. Thus, we reveals the existence of the size-effect law of fracture toughness of snow and we argue that the centrality of the Golden Ratio in the fracture properties of quasi-brittle materials. Consequently, by means of the proposed model, the Kic of large samples can be simply and rapidly predicted, without knowing the material performances but by testing prototypes of the lower dimensions.