Guidelines for the installation of median barriers presented in the AASHTO Roadside Design Guide have remained essentially unchanged for more than 30 years. In recent years, the need for improved guidance has prompted several states to reevaluate their guidelines and has also precipitated a nationwide research project administered by the Transportation Research Board. The objective of the study, on which this paper is based, was to develop improved guidelines for the use of median barriers on new and existing high-speed, multilane, divided highways in Texas. The purpose here is to present some modeling and benefitcost analysis results from that study, with a focus on the results from a particular data set developed under a cross-sectional withwithout study design. The highways of interest are those classified as Interstates, freeways, and expressways with four or more lanes and posted speed limits of 55 mph (88 km/h) or higher. The models employed to estimate median-related crash frequencies and severities, including the Poisson-gamma and ordered multinomial logit models as well as modeling results from a full Bayes estimation method, are presented. From the modeling results, a preliminary benefitcost analysis is described, in conjunction with some sensitivity analyses, for developing the guidelines for concrete and high-tension-cable barriers. A discussion of the limitations of this study and potential future extensions is provided.