The paper evaluates the possibility of modeling the heat transfer phenomena in a liquid-metal coolant using a light water test facility. It considers the natural circulation of the coolant in the upper plenum of the fast-neutron reactor. The sodium-cooled BN-1200 reactor was selected as the reactor installation to be modeled. The development of novel reactor designs must be based on the results of experimental studies. Some problems of modeling thermohydraulic processes in BN type reactors are studied by using sodium test facilities. Experimental studies of natural convection processes using light water test facilities can be considered as a good alternative to those using sodium test facilities. To validate the model, the similarity theory and the “black box” method were used and their principles and applicability were analyzed. Using the “black box” method makes it possible to avoid detailed modeling of such components as the reactor core and heat exchangers, replacing them by a simplified representation of these components to simulate the integral characteristics of the existing real life equipment. The paper considers the basic criteria which determine the similarity of the thermohydraulic processes under study. The governing criteria of similarity were estimated based on the fundamental differential equations of natural convection heat transfer. Based on these criteria, a set of dimensionless values was obtained which show the correlation between the model parameters and the characteristics of the reactor facility. Besides, generalized relationships were derived which can be used to estimate the scaling factors for calculating the key values of the reactor facility based on the model parameters. These relationships depend on the thermal-physics parameters of the working fluids, the geometrical scale value and the ratio of the thermal power of the model to that of the reactor facility, i.e., model-to-reactor thermal power ratio. The conditions under which it is possible to model sodium coolant by light water with adequate accuracy were analyzed. An example is given of the numerical values of the scaling factors for one of the reference light water test facilities. The paper uses the experience of a number of foreign researchers in this field, in particular, the accepted assumptions which do not result in serious loss in modeling accuracy. According to the available estimates, the assumptions used do not result in considerable losses in accuracy. Thus, the natural circulation of the sodium coolant in the upper plenum of the fast-neutron reactor can be simulated with adequate accuracy by using light water test facilities.