This work presents a new experimental approach to determine the reactivity levels of subcritical systems. The method employs the subcritical kinetic model developed by Gandini and Salvatores and it is based only on measured quantities such as counting rates of the detectors employed in the experiments and the parameters arising from the least squares fitting of the APSD (Auto Power Spectral Density) and CPSD (Cross Power Spectral Density). Detector efficiencies, quantity required in other procedures such as Neutron Source Multiplication (NSM) method, are not needed in the proposed method. The only hypothesis made in the method was the independence of the effective delayed neutron fraction and the prompt neutron generation time to the subcriticality level of the system. The proposed method was applied to measure the reactivity of several subcritical configurations of the IPEN/MB-01 reactor. Measurements of APSD and CPSD were performed in several degrees of subcriticality (up to around-7000 pcm). The spectral densities data were least squares fitted to get the prompt decay mode (α) and other quantities. Beside the startup source of the facility, an external neutron source of Am-Be was installed near the core in order to improve neutron counting statistics. The final experimental results are of good quality. The proposed experimental method shows clearly that the classical point kinetic theory cannot describe the measured reactivity. Instead, the reactivity inferred from this model follows closely the subcriticality index () for the source arrangements in the experiment. The agreement of the MCNP5 and GPT-TORT results, both with ENDF/B-VII.0 as the basic nuclear data library, when compared to the corresponding experimental ones was also good. vi SUMÁRIO BC1 e BC2 65.5% retirados-2 fontes de AmBe c