Interstate 80 (I-80) is a vital freight corridor running in the southern part of Wyoming. As the corridor being selected as one of the three sites for Connected Vehicle Pilot Deployment Program, a baseline assessment of existing countermeasures is vital. The 402-mile freeway has four weather-based Variable Speed Limit (VSL) corridors totaling 147-mile. VSL systems are countermeasures that aim to improve road safety by providing different operating speeds based on existing weather and road conditions. A before-after study with Empirical Bayes (EB) was utilized to develop Crash Modification Factors (CMF) for the VSL. Crash prediction models are essential to estimate the safety effectiveness of the implemented countermeasures. Parametric Negative Binomial (NB) and non-parametric Multivariate Adaptive Regression Splines (MARS) were used to develop crash models in the before period. Crash models were developed for total crashes, fatal and Injury crashes (FI), and Property Damage Only (PDO) crashes. Comparisons between CMFs obtained from NB model and MARS model were conducted. Analyses were performed using crashes occurred on the 402-miles of I-80 corridor, in Wyoming. Six years of crashes in the before period from 2003 to 2008 were utilized in the analysis, in addition to five years in the after period from 2012 to 2016. Results showed that the VSL reduced all the investigated crash types by different percentages ranging between 26% to 67%. Developed CMFs using NB and MARS showed comparable results. CMFs developed using the NB model might provide a quick and easy method for practitioners. Yet, it provides reliable results.