It is often the case that locality boundaries are not sharply delineated in our mental maps. This paper examines the level of uncertainty involved in perceiving qualitative boundaries of urban vs. rural localities. To measure location indeterminacy of locality, we begin with modeling locality as fuzzy region or also known as egg-yolk model which is composed of core, boundary, and exterior. The more a specific locality (e.g., Buffalo, Amherst) is geocoded within core, the more locality is location-determinant. 5460 fatal traffic accidents gathered from the Fatality Analysis Reporting System (FARS) in New York State from year 1996 to 2001 are classified into urban vs. rural cases, and are compared in terms of location indeterminacy. Location indeterminacy of rural localities are significantly higher than urban counterparts. It implies that perceiving boundaries of urban localities is less error prone than those of rural localities. The study shows that uncertainty is inherent in human cognition of geographic knowledge, and it poses challenges in geocoding imperfect data such as FARS.