Providing adequate design standards and capacity of speed-change lane is important for merging and diverging traffic to safely complete their maneuvers. Short speed-change lanes limit the ability of merging vehicles (/diverging vehicles) to reach equivalent mainline free-flow speed (/exit ramp speed). It also hampers proper gap seeking in the high-speed traffic stream. Hence, some speed-change (speed-up or slow-down) activities may occur on the mainline and disrupt mainline through traffic speed. While speed is the primary contributing factor of severe interchange crashes, many states have increased their urban interstate posted speed limit to 70 mph. However, not many researchers investigated the effect of increased posted speed limit on speed-change lane crashes. Therefore, the objectives of this dissertation are three-fold. The first objective is to identify critical factors through modeling and assessment of crash risk in merging speed-change lanes by freeway interchange type. The second objective is to identify critical factors and develop jurisdiction specific safety performance function (SPF) for freeway speed-change lanes, by severity. The third objective is to model and assess the effectiveness of increasing the posted speed limit on freeways on safety in the speed-change lanes. The findings are expected to help identify suitable geometric conditions (where) or what needs to be done if adequate speed-change lane length is not available.Freeway crash data was obtained from the Highway Safety Information System (HSIS), for the Mecklenburg County, North Carolina, from the year 2011 to the year 2015. Data for merging speed-change lanes along I-85, I-77, I-277, and I-485 in the Charlotte region, North Carolina were considered for identifying critical characteristics and modeling. Two different Multinomial Logit (MNL) models were developed for examining the relative risk of crash severity, by interchange type. Among the independent variables, freeway annual average daily traffic (AADT), speed-change lane length, ramp average daily traffic (RADT), speed difference between freeway and ramp, the number of lanes on the freeway and the ramp, weather condition, driver age, upstream and downstream ramp type, and upstream and downstream ramp distance were observed to be significant characteristics to model fatal and injury crash risk. The risk of crash severity is relatively higher for freeway sections with less than six lanes, single-lane merging ramps, and depends on closely located downstream ramp near a diamond interchange. In case of cloverleaf interchanges, adult-age drivers (41 – 55 years), single-lane merging ramps, and upstream ramp distance elevate crash risk in the speed-change lanes.For the SPF development, Negative Binomial distribution-based log-link model was considered as the crash data was observed to be over-dispersed. Data for 76 and 20 entry ramp speed-change lanes were used for modeling and validation, respectively. Traffic volume, speed difference between freeway and ramp, and ramp configuration were observed to be significant independent variables to estimate the number of total, property damage only (PDO), and injury type B and C entry ramp speed-change lane crashes. The predictability of the developed models was checked by computing the root mean squared error (RMSE) and the mean absolute error (MAE).The results obtained from the Empirical Bayes (EB) analysis show that increasing the posted speed limit on I-485 in Charlotte, North Carolina was not effective in reducing the number of crashes. The number of speed-change lanes experiencing an increase in the total number of crashes is greater than the number of speed-change lanes that experienced a decrease in the total number of crashes. Overall, the number of injury and PDO crashes was consistent in the after period for 10-30% of the speed-change lanes.