Corrosion-induced deterioration of reinforced concrete (RC) structures poses a significant challenge worldwide. Studies aimed at accurately predicting the flexural performance of corroded RC beams are essential for assessing the limit state of structural elements. This study proposes predictive models for the load-bearing capacity of corroded slender RC beams using data from 145 beams tested in previous investigations. The database was analyzed using an artificial neural network (ANN) and an improved beam section model to identify critical parameters and develop a semi-empirical formula. The results indicate that the ANN model can effectively predict the ultimate flexural strength of beams with corroded longitudinal reinforcement, achieving an R-squared value of 0.9882. Parameter importance analysis enabled the development of a semi-empirical formula that can be conveniently applied by engineers. The beam section analysis-based formula also provides highly accurate predictions of load-bearing capacity, with an R-squared value of 0.9688. A comparison with previous formulas shows that the proposed models yield superior results.