The relative risk (RR) of cases to controls over a region can be expressed as the ratio of their probability density functions. The k-th Nearest Neighbor (kNN) estimator of the RR is based on the non-parametric kNN density esti- mation. For large samples, its distribution can be approximated by an incomplete beta function. We will present asymptotical properties of the kNN RR estimator, as well as numerical work supporting the theoretical findings. A comparative analysis shows that for large samples the power of the kNN method is comparable to that of the kernel RR method. However, the kernel approach for assessing the significance of the estimated RR is completely computationally based. In contrast, the kNN method has theoretical justification and provides a straightforward closed form for the bounds of the rejec- tion region. The kNN RR method has been implemented as an exploratory tool in the Kentucky Occupational Injury Surveillance system. A study using the kNN RR method to identify geographical areas with elevated RR for at- fault commercial truck collisions at night in Kentucky will be presented.