The detection of breathing cracks in in-service structures is an intractable issue due to the distinctive dynamic behavior, typically the opening and closing of the crack subjected to cyclic loading. The existing methods to identify breathing cracks generally rely on the hypothesis of the crack-induced nonlinear features, mostly represented by super-harmonics. Nevertheless, an in-service structure even free of cracking somehow has nonlinear components in material, geometry, boundary etc., such that it tends to present inherent nonlinear features, i.e., nonlinearities prior to cracking, which could impair the effectiveness of super harmonics-like methods to depict breathing cracks. In such a circumstance, a sophisticated method that can characterize breathing cracks-induced nonlinearity by accommodating inherent nonlinear of the structure is desired. Herein, a novel method, termed DISTINCT (Difference in Split Temporal signal with Inherent Nonlinear Components Tolerated) method, is proposed. This method is formulated by splitting a whole response of a structure bearing breathing cracks into positive and negative response components. Each response component largely reflects structural dynamic behavior with regard to the crack-opening state or crack-closing state. From positive or negative response components, the dynamic property such as the frequency spectrum arising from the Fast Fourier Transform, corresponding to crack-opening state or crack-closing state, can be derived. To this end, the differences between the two frequency spectra relatively purely manifests the characteristic of breathing cracks including their presence and severity. The feasibility of the DISTINCT method is numerically verified via the identification of breathing crack in a cantilever beam and a shell, with emphasis on its stronger noise resistance than the traditional super-harmonics-based method. Furthermore, the method’s effectiveness is experimentally validated via the inspection of breathing cracks in a scaled arch dam model subject to excitations induced by artificial seismic accelerations. The results show that the proposed DISTINCT method can quantify the severities of four levels of damage comprising breathing cracks by tolerating inherent nonlinearity of the structure. The advantages of the DISTINCT lie in its tolerance to structural inherent nonlinearities and requiring no numerical or physical benchmarks nor any prior knowledge of the material properties and boundary conditions in assessing breathing cracks of structures.