Corrosion is one of the most typical form of durability degradation of steel bridge and the corrosion morphology takes significant effect on the strain distribution of the structural surface. In order to investigate the characteristics of steel corrosion morphology as well as the effect of the corrosion on the structural ultra-low cycle fatigue(ULCF) performance, this paper takes the steel kind of Q345 as the study case, conducting 2D and 3D profile measurements on four groups of steel plate samples of different corrosion conditions. Quantitative analysis on 2D profile data were performed with the methods of surface roughness, fractal dimension, and power spectrum density function. Meanwhile, precise finite element models of corroded specimens were established on the basis of 3D profile data. On the basis of the improved cyclic void growth model (CVGM), the ULCF cracking theory is taken into deep investigation, and the issue on the key factors leading to corrosion samples cracking are discussed. The research results indicate that: the corrosion rates of the artificial acceleration corrosion tests are significantly affected by the thickness of the rust layers. However, the surface roughness, fractal dimension, and power spectrum density function are unable to differentiate the corrosion degree. According to the numerical calculation results, the stress triaxiality and equivalent plastic strain on the corrosion pits are much higher than that of the other zones, such that corrosion pits have critical impact on the steel ULCF cracking.