Prediction of traffic accidents has an important meaning in improving of road safety. Models of traffic accidents prediction can help for finding accident causes and to reduce the number of accidents. The purpose of this paper is to build a fuzzy logic model for prediction of traffic accidents in Albania. In this model are used six input parameters: annual average daily traffic (AADT), road width (rw), speed (sp) defined as the distance per unit of time, number of minor access (ma) along the street over the length of the street in kilometers, road surface condition as percentage (pm), and the percentile of sign per kilometers of road (sj). The output of the model is the annual all accidents (AAA) that is defined as the number of all accidents occurring on the road in a defined time interval of a day per kilometer length of road. This model is applied for seven roads in Albania. The results are discussed and taken the main conclusions of this paper. Here fuzzy logic toolbox of MATLAB is used.