The Variable Message Sign (VMS) has been deployed in Minnesota since 1960s. The evaluations for the effectiveness of the signs are critical for their appropriate installation and deployment. In this study, five VMS devices were selected along Interstate-94 between downtown Minneapolis and St. Paul. The data collection period ranged from January 2006 to December 2012. First, a linear model was used to investigate the effect of VMS on the speed changes, in the corresponding impact regions. The analysis was conducted using two scenarios with eight different conditions. The results revealed that the speed changes, which were influenced by the display of VMS messages, were within 2.0 miles per hour (mph) for all the conditions. Moreover, adverse weather, and times of day were unlikely to affect the model results. Second, a 2×2 contingency table, and a logistic regression model were used to explore the association between the deployment of VMS and the crash occurrence. Odds ratios for the probabilities of crashes under the impact of VMS were estimated. The results indicated that the deployment of VMS messages was not likely a risk factor for crash occurrence. The estimated odds ratios for both warning and informative message types were not significantly different from 1.0.