Identifying and treating any inconsistency on a highway can significantly improve its safety performance. Potential measures used to evaluate the geometric design consistency of highways are operating speed, vehicle stability, alignment indices and driver workload. Although previously developed models are useful, consistency measures such as vehicle stability and driver workload are not considered. Therefore, the objective of this study is to develop accident prediction models in order to investigate and predict the impact of geometric design consistency measures on the safety of two-lane rural highways. A combination of geometric, traffic and five-year accident database at 44 sections along Bahir Dar–Debre Markos roadway has been used to investigate the impact of several design consistency measures on road safety. Inconsistent sections have been identified based on speed and vehicle stability criterion. About 36.0% and 25.0% of the roadway sections are evaluated as inconsistent sections based on speed and vehicle stability criteria respectively. Moreover, about 25% of inconsistent sections are also identified as black spot locations. Eight individual models representing the impact of each consistency measures on road safety and two overall models (Model-A and B) relating several consistency measures and road safety have been developed. These models can be used as a quantitative tool to evaluate the impact of geometric design consistency on road safety. The generalized linear regression approach has been used for model development. To verify the accuracy of Model-A and B, validation steps have been performed using equivalent data of Debre Markos to Dejen roadway sections. All estimated errors for both Model-A and B are less than the expected error (< 0.05). Therefore, the two crash prediction models are valid enough to represent the observed crash data. Finally, implementable countermeasures such as parabolic speed bumps and transition curves were designed to reduce the frequency and severity of traffic accident at inconsistent sections identified based on speed and vehicle stability criteria respectively. In addition, black spot sections were prioritized based on their weighted severity value to perform detail safety diagnosis.