One task of traffic safety engineers is the identification of high crash locations and selection of appropriate highway treatments to reduce the number of crashes. This process relies on the availability of accurate information on crash reduction factors of various treatments. Currently, most agencies rely on information dating back to the 1960s. It is necessary to update and reassess these factors using new data and new evaluation methods. A before-and-after study is currently being conducted using empirical Bayesian methods to estimate crash reduction factors for modern conditions on two-lane rural highways. The results of the second phase of the study are reported, which aimed to evaluate the safety benefits of intersection approach realignment. Furthermore, an analysis of variance model is used to identify extra benefits of comprehensive treatments. The improvements studied appeared to reduce the total number of crashes, but the effect on type of crashes was different. Also, combining realignment with adding a left-turn lane or traffic signal does not appear to offer significant additional benefits in crash reduction.