The aim of this research is to determine the effect of unsignalised intersection geometry on the rates of the various types of accidents occurring at unsignalised intersections. A literature review has identified that there is little consistency between the results of previous studies. Some studies found that particular parameters had an opposite effect to what was expected. With this in mind, the research identified reasons for these results and developed two basic approaches to mitigate some of the problems with multi-factor type studies. These approaches are 'maximise the efficiency of data collection' and 'develop techniques for analysing less than perfect data'. A database consisting of 206 unsignalised intersection sites from throughout Queensland was used for analysis. The outcome of this research confirms the validity of several of the current design standards for unsignalised intersections, in addition to identifying new engineering procedures.