In conducting the airframe structural design the strain diagrams must be predetermined; for thin-walled airframes these diagrams are chosen frequently to be "idealized", e.g., diagrams without hardening or with linear hardening, diagrams of ideally elastic material, and others. As to calculation, it is carried out on rupturing loads whereat the stresses and strains are interrelated by nonlinear dependences. The work of material beyond the limits of proportionality is taken into account by using different methods of iteration such as the method of reduction coefficients, the method of secant moduli, the method of fictitious loads, and so on. The strain diagrams s -e and t -g taken usually into consideration can be obtained by statistical processing of a large number of experimental data on specimens tested at simple loads. As to the elements of real thin-walled airframes, they work under different conditions. In this connection, the question arises of whether the diagrams constructed for specimens can be applied to the elements of a real thin-walled structure. It is necessary thus to determine diagrams (or to refine the diagrams known only approximately). In this paper we suggest to construct these diagrams using the data of tests as the base. By the notion "tests" we shall mean the ground-based full-scale experiments aimed at determination of the thin-walled airframe strength.