Abstract: Objectives Existing studies rarely consider the impact of the uncertainty in corrosion distribution on the load-bearing capacity of hull structures, as well as the residual ultimate strength of aging hull structures under vertical bending, which to some extent limits the reliability and applicability of related assessment methods. Methods Based on empirical pitting depth data of hull plates from existing studies, this approach introduces a depth distribution model of actual corrosion pits on hull plates, enabling a stochastic pitting corrosion modeling method. It establishes a numerical model to evaluate the residual ultimate strength of corroded large-opened box girders. Developed as a secondary Python application within ABAQUS, this method allows precise control over multiple corrosion parameters, such as relative corrosion volume, ensuring the generation of pit depth distributions aligned with target distributions. The entire numerical simulation process integrates automation, culminating in a nonlinear finite element analysis that parametrically evaluates corroded large-opened box girders under vertical bending moments. Results The results indicate that, under the same corrosion volume loss with a regulated pitting depth distribution, the difference between the maximum and minimum average ultimate strength among the four groups of models with varying pitting radii is 0.89%. When the relative pitting area or relative pitting depth decreases linearly, the mean value of the ultimate strength reduction factor θ declines linearly, reaching a minimum of 0.83 and 0.85, respectively. With a Weibull distribution of pitting depth and a relative pitting volume loss up to 5%, the ultimate strength of the box girder decreases by a maximum of 16.7% and a minimum of 12.6%, attributed to the spatial variability of pitting distribution. Additionally, a probabilistic distribution model for the ultimate strength under varying relative corrosion volumes is fitted. Conclusions A method for constructing corroded box girder models that generate target pitting depth distribution characteristics is proposed, achieving fully automated integration in numerical simulations. This approach, which combines the uncertainty of pitting spatial distribution with nonlinear finite element analysis, provides a more accurate prediction of the residual ultimate strength of corroded hull structures under vertical bending moments.