This research was to develop a Traffic Accident Analysis System (TAAS) to aid in the identification of accident black spots and develop a statistical model to predict traffic accident severity. TAAS was developed as a set of python tools and deployed as a toolbox in ArcGIS© 10.X. There were all together more than 252,251 traffic accidents (from 2008-2014) reported in Sri Lanka. TAAS consists of data for 20,041 traffic accidents reported in the Southern province of Sri Lanka over eight years (2008-2014). All relevant attributes of traffic accidents in the possession of the traffic police were included in TAAS. (Traffic Police Statistics in Sri Lanka 2014). According to the World Health Organization (WHO) [1], more than 1.3 million people die each year in traffic accidents and more than 50 million are injured worldwide (WHO 2012). Sri Lanka traffic police analyse traffic accidents through a software called MAAP. The collected data are not properly used for analysis because it cannot be done in a user-friendly manner. As a solution to this weakness, a GIS-based accident analysis system [2] which links a great volume of accidents was developed. As for the second objective of this study a logistic regression model was developed to predict the traffic accident severity. 2,802 serious and fatal traffic accidents were used in the model. Out of eight independent variables used, three were found significantly associated with traffic accident severity: „Time‟, „Road Surface Condition‟ and „Days of Week‟.