The objectives of this study investigating Changeable Message Signs (CMS) were to determine whether or not CMS messages really work, whether or not they cause traffic slow downs, and whether or not they have an impact on traffic flow. The participants were 120 licensed drivers from three age groups-18-24, 32-47, and 55-65 years old. Two experiments were conducted in a fully-interactive, PC-based STISIM driving simulator. Experiment One investigated the effectiveness of the following message, CRASH/AT WYOMING AVE/USE THOMPSON EXIT. In Experiment Two, the final CMS message was: AMBER ALERT/RED FORD TRUCK/MN LIC# SLM 509. Results. In Experiment Two, only 8.3% of the participants had Excellent AMBER Recall Scores, while 51.7% had Good scores. Gender significantly affected the AMBER Recall Scores-there were more females than males in the Excellent Category. A greater proportion of those who knew what AMBER Alert meant were in the Excellent and Good Categories. 21.7% of the participants slowed down by at least 2 mph. Whether or not traffic delays will result from drivers slowing to read AMBER Alerts in real life will depend on the extent of the slow downs and on current traffic density. In Experiment One, 55.8% of the participants took the Thompson Exit after seeing the Thompson Exit Message. Of the 53 participants who did not take the exit (1) 35.9% ignored the CMS message because they did not think that it applied to them; (2) 35.9% did not understand the CMS message; and (3) 22.5% did not notice the message. (It is not known why 5.7% of the 53 did not take the exit.) Changes to the wording of the messages are recommended.