The study develops a methodology by using the elements of GIS to understand the intra-city traffic accidents. Four-year accident data for the city of Manhattan, New York was analyzed using functionalities of GIS and further the spatial- temporal patterns were identified based on 2-D and 3-D visualization of the data. Moreover, the last section deals with the Bayesian approach based on which the high risk road segments were analyzed using stable crash rates of the city. Accordingly, this approach provided with a ranking of the high risk road segments so as that preventive and advanced safety measures can be taken and public can be diverted for taking safer routes. Results demonstrate the approach to be helpful in estimating the relative accident prone routes, eliminating the instability of estimates using the two important variables population number and road-junction density. The 2-D and 3-D visualization show risky zones where safety measures are required to be implemented.