Because of the coupling of environment corrosion and wind-induced vibration, the fatigue properties of angles have degenerated during the service period. However, the fatigue testing data is scarcity owing to time-consuming and costly. In this work, based on the thinking of saving cost and improving precision, an improved regression method for estimating S-N curves of corrosion fatigue was proposed combining with the single-point likelihood method (SPLM) and optimization method of correlation coefficient. And the implementation algorithm for the improved SPLM was also proposed. According to this improved method, a t-P-S-N surface expression of corrosion fatigue was presented. Meanwhile, three types of 18 Q345 equal angles in different corrosion time were studied by the corrosion fatigue test. Through the comparison of results, corrosion fatigue failure modes of specimens were observed and the variation law of P-S-N curves within different corrosion time was obtained. Combining this improved SPLM, the t-P-S-N surface model of Q345 equal angles was proposed. At last, the variation law of Q345 equal angles’ fatigue life was studied by the proposed t-P-S-N surface model. It can be found that: 1）The fatigue life was decreasing sharply with increasing corrosion time under the same load. 2）The effect of corrosion on the fatigue life was becoming more and more serious with decreasing load in the same corrosion time.