The majority of fatal and serious crashes occur on rural two lane roads. A number of research studies have attempted to quantify the effect of various road features on rural road crash rates. Policymakers are particularly interested in where they should invest their money to drive down crashes, e.g. should they invest in rural realignments or road widening. Traffic engineers want to try and diagnose the safety problems on a particular road. This is difficult on low volume roads where crash numbers are low and crash patterns are not obvious. No existing crash prediction models allow analysts to look at all the key variables and their interactions. The purpose of this research is to develop a new crash prediction model with all the key variables. A sample set of the entire New Zealand two-lane State Highway network has been compiled. This paper covers the scoping and pilot phases of this study and the work that has been completed to date on the main study model. In addition, the paper discusses ongoing work in New Zealand to adapt recent road safety developments from the US, such as the Interactive Highway Safety Design Model (IHSDM). (a) For the covering entry of this conference, please see ITRD abstract no. E220164.