A unifying theory is presented for the process of heave in freezing soils. Out of the Cold Regions Research and Engineering Laboratory (CRREL) school of D.M. Anderson came the concept of the segregation potential. Out of the Cornell school of R.D. Miller came the model for the heave rate. Here ideas from both schools are put on a fundamental thermodynamic footing, leading to the definition of a new heave index. Both schools use the temperature gradient in the frozen fringe as the driving force for heave. We argue and demonstrate that this choice leads to erratic results. The driving force should be the temperature gradient over the complete layer of soil that is at sub-zero (°C) temperature, that is, the combined frozen zone plus the frozen fringe. The value of the heave index is completely dominated by the unsaturated hydraulic conductivity function of both the unfrozen soil below the frozen fringe and the soil layer at sub-zero temperature.