Abstract: To accurately predict the evolution of the uniaxial compressive strength of sandstone subjected to chemical erosion by solutions with different pH values during wet-dry cycling. Based on 166 sets of sandstone wet-dry cycling test data, gene expression programming (GEP) was used to construct a prediction model for uniaxial compressive strength that incorporates environmental factors and material properties. The effects of six variables on sandstone strength degradation and parameter sensitivity were investigated. The results show that: (1) The established GEP model demonstrates excellent predictive performance, with the coefficient of determination (R²) for both the training set and the test set exceeding 0.92, which is significantly better than that of six classical benchmark models; (2) The single-cycle duration exerts a three-stage degradation effect on sandstone strength: strength attenuation is pronounced within 10 wet-dry cycles, and the sandstone strength decreases linearly as the pH value decreases; (3) When the SiO₂ content of sandstone increases from 10% to 80%, its compressive strength increases by 14.5%; as the dry density increases from 2.06 g/cm³ to 2.60 g/cm³, the sandstone strength increases from 22.69 MPa to 52.16 MPa; in addition, when the initial moisture content is within 6%, strength attenuates relatively rapidly. (4) Sensitivity analysis reveals that the influence of environmental factors (with a contribution of 68.3%) is significantly greater than that of material properties (31.7%), among which the number of wet-dry cycles (27.9%) and single-cycle duration (25.1%) play dominant roles. This GEP model converts complex nonlinear relationships into interpretable mathematical expressions, with both predictive accuracy and physical interpretability. It thus provides a reference for analyzing the strength degradation mechanism of sandstone under wet-dry cycles.