This project will develop a simple tool to assist State DOTs in predicting the future retroreflectivity performance of pavement markings for a period of at least three years, without having to perform costly and time-consuming retroreflectivity measurements on a regular basis. The research team will use the National Transportation Product Evaluation Program (NTPEP) data mine to collect retroreflectivity data along with the corresponding key variables for at least 5,000 pavement-marking lines. About 90% of the collected data will be used to train six different supervised machine-learning models to predict retroreflectivity at various time intervals (1, 2, 3, 11, 12, 15, 21, 24, 27, 33, and 36 months), based on the initial retroreflectivity and the corresponding variables. The remaining about 10% collected data will be used in testing the developed machine-learning models. The test results for the six machine-learning models will be compared to select the model providing optimum accuracy. Stage 2 will involve field tests in which two test decks will be constructed in the southern and the northern U.S to monitor the retroreflectivity of pavement marking materials installed on road sections. The deck design will consider variables such as different pavement marking materials, pavement surface, traffic levels, etc. The test deck measurements will be used to calibrate the selected machine-learning model. The calibration process will also utilize data from the “Measure across America Project.” A simple tool will then be developed for the calibrated machine-learning model that will allow state agencies to use the proposed machine-learning model without the need for coding software (such as MATLAB or Python).