The relationship between unfrozen water content and temperature in frozen soils, which is referred to as the phase composition curve (PCC), is a fundamental relationship in cold regions engineering. In a previous study, the authors succeeded in developing a physical description and a physically-based equation for the PCC, which overcomes the limitations of the existing empirical approaches. Here, the authors propose a physico-empirical approach to predict the parameters in this equation to facilitate the calculation of the PCC in practice. An accurate prediction of the PCC will only need simple soil index properties and one measured data point for constraint. In this approach, the four parameters in the PCC equation are first calculated from soil index properties using accepted formulas. Two selected parameters are then adjusted by a curve fitting process using the measured data point. A criterion was suggested for obtaining the best point. This new approach was implemented using a computer program to automate the process. Validations with data from several soils indicated that the approach offers consistent and accurate predictions of PCCs when used with Zapata's model for plastic soils and with the Mechanistic–Empirical Pavement Design Guide (MEPDG) model for non-plastic soils. This study thus bridges an important gap between the theory and application of PCCs.