Key nuclear inputs for the astrophysical r-process simulations are the weak interaction rates. Consequently, the accuracy of these inputs directly affects the reliability of nucleosynthesis modeling. Majority of the stellar rates, used in simulation studies, are calculated invoking the Brink-Axel (BA) hypothesis. The BA hypothesis assumes that the strength functions of all parent excited states are the same as for the ground state, only shifted in energies. However, BA hypothesis has to be tested against microscopically calculated state-by-state rates. In this project we study the impact of the BA hypothesis on calculated stellar β--decay and electron capture rates. Our investigation include both Unique First Forbidden (U1F) and allowed transitions for 106 neutron-rich trans-iron nuclei ([27, 77] ≤ [Z, A] ≤ [82, 208]). The calculations were performed using the deformed proton-neutron quasi-particle random-phase approximation (pn-QRPA) model with a simple plus quadrupole separable and schematic interaction. Waiting-point and several key r-process nuclei lie within the considered mass region of the nuclear chart. We computed electron capture and β--decay rates using two different prescriptions for strength functions. One was based by invoking BA hypothesis and the other was the state-by-state calculation of strength functions, under stellar density and temperature conditions ([10, 1] ≤ [ρYe(g/cm3), T(GK)] ≤ [1011, 30]). Our results show that BA hypothesis invoked U1F β-− rates are overestimated by 4–5 orders of magnitude as compared to microscopic rates. For capture rates, more than 2 orders of magnitude difference was noted when applying BA hypothesis. It was concluded that the BA hypothesis is not a reliable approximation, especially for the β--decay forbidden transitions.