Rocks under compressional stress conditions are subject to long‐term creep deformation. From first principles we develop a simple micromechanical model of creep in rocks under compressional stress that combines microscopic fracturing and pressure solution. This model was then upscaled by a statistical mechanical approach to predict strain rate at core and reservoir scale. The model uses no fitting parameter and has few input parameters: effective stress, temperature, water saturation porosity, and material parameters. Material parameters are porosity, pore size distribution, Young's modulus, interfacial energy of wet calcite, the dissolution, and precipitation rates of calcite, and the diffusion rate of calcium carbonate, all of which are independently measurable without performing any type of deformation or creep test. Existing long‐term creep experiments were used to test the model which successfully predicts the magnitude of the resulting strain rate under very different effective stress, temperature, and water saturation conditions. The model was used to predict the observed compaction of a producing chalk reservoir.