The subject of the study was to determine the reliable characteristics of the traffic flow variability for the purpose of estimating the annual average daily traffic, including the year of the forecast. In order to achieve this objective, there was performed traffic volume data analysis between 2000-2010. These studies helped to identify traffic flow variability at different periods of time, its determinants, as well as their impact on traffic flow characteristics which are relevant for estimating the AADT.The research presented: division of the road networks into uniforms traffic sections (cluster analysis, functional/geographic classification); representative traffic volume profiles (by definition); model relationships between time-related characteristics of traffic volume variability and AADT (regression analysis, artificial neural network); model intended to complete missing traffic data (SARIMA). Conclusions from the conducted research create theoretical and empirical basis to enhance the quality and reliability of determining AADT and, consequently, the quality of road solutions.