The coefficient of thermal expansion (CTE) is one of the material properties of concrete that has the largest impact on rigid pavement performance. Concrete CTE is typically measured in the laboratory, under saturated conditions, or estimated on the basis of the mix constituents, past experience, or both. Whichever method is used, the mechanistic-empirical design of concrete pavements traditionally assumes a constant value for this material property. This assumption has important consequences in relation to predicting thermal deformations and stresses since the CTE of concrete actually changes with the concrete’s internal moisture conditions. The experimental data presented in this study show that this assumption, together with the way CTE is measured in the laboratory, leads to systematic underestimates of thermal deformations and stresses in concrete pavements. The experimental data come from six concrete overlays of asphalt pavements that were instrumented with thermocouples, relative humidity sensors, and vibrating wire strain gauges to measure the expansion/contraction and bending of the slabs because of temperature and moisture-related actions. The apparent CTE of the overlay slabs reached values up to 65% larger than the CTE measured in the laboratory under saturated conditions. Using finite element method modeling, it was determined that thermal stresses were up to 70% larger than predicted using the saturated CTE.