Thermo-hydro-mechanical–chemical (THMC) coupled processes that have resulted from long term past climate changes and glaciation cycles in the sedimentry rocks of southern Ontario are investigated. A conceptual numerical model has been developed to solve four coupled partial differential equations (PDEs), which represent hydraulic, thermal, mechanical and chemical processes. The finite element method is used to solve the PDEs under transient surface boundary conditions imposed by past glaciation cycles to predict the hydraulic, mechanical, thermal and geochemical responses of the geological system. The results show that past glaciations have a significant impact on the hydraulic gradient and pressure, vertical effective stress and salinity profiles, and a limited effect on the permafrost depth. The predicted results show good agreement when compared with the field data for the total dissolved solid, rock strength and quality. The results show relatively good agreement with the anomalous pore water pressure profile in the field. The modeling results indicate that the infiltration depth of glacial melted water is less than 300m, and are consistent with the field observation of total dissolved solids. At the level of a deep geological repository (DGR) for low and intermediate level radioactive wastes being planned in these rock formations, a safety factor of 6.9 is predicted against failure by using Hoek–Brown failure envelopes, while a low safety factor of 0.83 is predicted at the shallower level of the Silurian (Salina) formation. It is found that solute transport at the middle and upper Ordovician formations are diffusion dominated at depths of 300m or more, and controlled by diffusion–advection above 300m. Based on the results obtained, the modeling of a past glaciation can be used with reasonable confidence in predicting the impact of future glaciations related to the long term safety and stability of the proposed DGR in the sedimentry formation. However, for site specific conditions, THMC modeling is very sensitive to material properties, and sensitivity analysis is required for future model development.