Ship hull structural maintenance practices follow a preventive approach that has remained virtually unchanged over recent decades. Recently, Structural Health Monitoring (SHM) has emerged as a viable option to transition towards condition-based maintenance (CBM). This work aims to develop an SHM system for corrosion-induced thickness loss (CITL) estimation in ship hull structures that uses strain data. For this purpose, a simple rectangular plate experiencing uniform corrosion was considered as a reference structural element, which was subjected to stochastic loading. Strain response data were produced through a high-fidelity Finite Element (FE) model to simulate real-world measurements. CITL was estimated within a Bayesian framework employing surrogate models to enable a computationally efficient solution. Results indicated that strain can be used effectively to indirectly monitor CITL and that the efficiency of the proposed approach depends on the level of domain expertise introduced to the model via the selection of Bayesian priors.