The characterization of steel fiber concrete in a flexural test or tensile uniaxial test requires greater control of loading rate by the speed applied in the testmachine, to be able to report more accurately the behavior of displacementsor crack openings in micro-scale possible. In general, in the pre-dimensioningprocess, it is often discarded the characterization tests and the main standardsdo not provide theoretical design recommendations concerning the residualstrengths of fibrous concrete. Some empirical models in the literature attempt anapproximation of the theoretical design values, but they present greater dispersion in the results. This article proposes empirical equations for predicting theresidual strengths of the steel fiber reinforced concrete (SFRC), for small fibervolume fractions (approx. < 0.8%). Experimentally notched beams subjected tothree-point bending tests were collected, with the concrete matrix composed ofhook-end steel fiber and the experimental residual strengths presenting the softening behavior in the load-opening ratio curve of the crack. The experimentalresidual strengths were compared with other empirical models found in the literature. The results showed that the maximum stress of the steel fiber pulloutmodel established satisfactory relationships with the residual stresses in flexure.The empirical proposed residual strength model presented lower variability ofdispersion around 10%, the mean absolute percentage error of 8%, compared toother usual empirical models and had the same material class with the experimental results, proving to be a viable alternative for pre-design and dosageoptimization for SFRC.