UDC 539.376 The problem of describing thermoviscoplastic deformation of materials arises in modeling the behavior of structural elements under intensive loading and elevated temperature conditions, which in particular are characteristic of various industrial treatments such as welding, forging, etc. [3, 14, 18]. A rather complete review of results obtained within approaches to modeling the indicated processes is given in [5, 6]. Most of them to some degree use the concept of the yield surface and representation of the strain rate as the sum of terms corresponding to the processes of elastic and instantaneous plastic deformation and creep deformation. At the same time, it is commonly accepted that plastic deformation essentially is a time-dependent process. Attempts to develop a theory of plasticity taking into account the time and temperature dependence led to the creation of so-called unified theories. In these theories, all the characteristics of inelastic deformation are represented in a single inelastic strain term. In the evolution equation for inelastic deformation, internal variables appear whose number depends both on the hardening mechanisms (isotropic, directional) and the complexity of the loading histories under consideration. In such a formulation, the model does not explicitly contain a yield condition and related loading and unloading conditions. Thus the dependence of plastic flow on the strain rate under monotonic and cyclic loading, and also creep and stress relaxation, are determined from a single system of equations. Unified theories include those in [4, 9, 10, 15, 16]. The results of experimental study of these theories for the case of a uniaxial stress state are presented in [2, 8, 10]. It has been established that the given theories describe well the substantially nonlinear, temperature-dependent behavior of metals under complex thermomechanical loading. References [13, 19] are devoted to generalization to the case of multiaxial states. A number of papers are also available which make it possible to give a positive empirical evaluation based on comparison of finite-element calculations using the indicated models and the experimental data in problems of stress concentration and crack propagation [17], and also residual weld stresses [18]. Comparison of the major models in this class [2] shows that quantitatively and qualitatively, the thermoviscoplastic behavior of metals is described most exactly by the Bodner-Partom model [10, 11]. The paper [8] is devoted to development of a flow chart for determining the parameters of this model. In approaching this problem, we should bear in mind two facts reflecting the specifics of the model: first of all, the responsibility for the characteristic aspects of the inelastic behavior, such as the yield stress and directly several parameters of the model; secondly, the sensitivity of the model to the quality of the experimental data and the values of a number of its parameters. From this standpoint, the flow chart proposed in [8] involves assumptions and procedures (in particular, the procedure for differentiation of the experimentally obtained so-called ~,-curves) lead~g to errors in determination of the model parameters. In this paper, we develop a correct procedure for determining the desired parameters of the Bodner-Partom theory of thermoviscoplasticity, which we illustrate using the example of the AMG-6 aluminum alloy. 1. Equations of the Bodner-Partom Model. The system of constitutive equations for the Bodner-Partom model