Local calibration of the punchout model in the Pavement ME software is a vital step in achieving performance predictability for the design of Continuously Reinforced Concrete (CRC) pavement. In Oklahoma, there was only limited performance data available in the General Pavement Studies (GPS) database for CRC pavement. This set of circumstances required a different approach as to the type of data used for calibration. The type of data originally utilized in NCHRP 1-37A essentially represented visually evident damage that is clearly observable at the surface of the pavement structure. Non-observable damage however is actually of greater value as a source of calibration data since it represents the deteriorative conditions that lead to the visual manifestation of the damage process. Since visually validated distress is the end result of the distress cycle the traffic level associated with it is often subject to a considerable amount of error. In this regard, non-observable data such as erosion damage is shown to be a good indicator of and a substitute for actual punchout data since it represents the deterioration of the slab subbase interface that has be found to closely aligned with the punchout process. The amount of erosion is evaluated based on FWD data and is shown it to be a reliable way to determine the calibration coefficients for the punchout model. This paper proposes an approach for calibrating local coefficients for CRC pavements based on non-observable performance data. The main process of this methodology requires estimating erosion percentage damage using Falling Weight Deflectometer data (FWD), determining the percentage of punchout from the Long-Term Performance Program (LTPP) records, and establishing the relationship between both components to estimate the existing punchout distresses. This relationship can be used to calculate the actual damage including erosion damage and to calibrate the local coefficients used in the pavement ME punchout model. This methodology was carried out on one section from Oklahoma and one section from Texas in order to validate its applicability and conclude on the pavement ME punchout model and its ability to predict punchout distress in the field.