Traditional statistical methods are more appropriate for investigation of influences of internal factors – for localized objects investigation. In most cases possibility of traditional statistical conceptions and methods for investigation of real socio-economical object is bounded. The method scan statistics allows investigating complex systems both as a whole and by their separate parts and thus allows managing the investigated process more effectively. Scan statistics allows to investigate the socio-economical problems having extremely complex, i.e. sinenergic structure of interrelations (structure of the open systems). The opportunity of use scan statistic methodology for research of accident events on roads of Latvia is shown in paper. The traditional approach is research of process of occurrence accident events on all roads of territory of Latvia as a whole depending on time that does not allow defining clusters of accident events on roads of Latvia. In difference from the traditional approach in work the technique of application scan statistic methodology for research of process of occurrence accident events in space, namely on separate regions of Latvia is considered. It allows defining objectively existing clusters accident events. In work stochastic modelling p-value by a method of Monte-Carlo for definition accident black spots on roads of Latvia is used. The given information allows to improve a control system of the organization of movement of transport on roads of Latvia and to make of the decision for perfection as global and local infrastructure of roads of Latvia. The researchers seek to use scan statistic to monitor future road accidents data or to design a Latvian roads management system with the purpose to minimize the number of accident events.