It is well documented that motor vehicle crashes are a public safety concern. However, traditional approaches do not always lend themselves to addressing the complete extent of this "safety problem". Identifying the extent of the "safety problem" is an important step in optimizing safety fund allocation and analyzing horizontal curve safety. This study investigates the allocation of safety expenditures in Iowa, relative to crash data. The matching of crash data with safety expenditures suggests the shift of funds from the high crash density, state system to facilities on the low density, local system. However, the redistribution of funding should also consider factors such as crash density and benefit cost. Furthermore, because some crashes are too widely distributed to be identified using traditional high crash location methodology; a balance of systematic and high crash location methods should be considered. Ultimately, the optimum balance of safety resources should reduce the most possible fatal and serious injury crashes. This study also investigated a systematic method for identifying and estimating geometric parameters on horizontal curves. A validation of this method showed that as horizontal curve radius decrease, sensitivity to errors in the estimated curve radius increase. Although some large errors associated with the estimated curve radius were found, predicted crash frequency for all curves was found to be no more than twenty percent different than the actual predicted crash frequency. Lastly, safety performance functions created for the horizontal curve database did not yield a concrete correlation between curve radius and crash frequency. Because of the random nature of fatal and major injury crashes, care is advised when creating crash models for these crashes.