The era of intensive construction of new roads is behind us, so road agencies are now focused on maintaining and preserving existing pavement surfaces. As they are faced with limited founds for maintenance it is important to utilize the money by selecting the best maintenance strategy. Selection of appropriate maintenance strategy is a complex task which includes factors such as current condition of the pavement, road classification, traffic volume and type of pavement distress. These factors can be automated and implemented in pavement management systems to achieve standardised approach to road pavement assessment and management. One of the key components of a pavement management systems are pavement performance prediction models which simulate pavement deterioration process and forecast its condition over time, one of such model is artificial neural network. This paper analyzes the possibility of using artificial neural networks in pavement management systems for evaluation of existing pavement condition and its possible application for defining the maintenance strategy of state roads. Backpropagation algorithm was applied on 481, 3 km of state road in Osijek-Baranja County which represents 7% of total length of national road network in Croatia. Obtained results indicated that artificial neural networks can be used for optimization of maintenance or rehabilitation strategies as well as assessment of pavement condition at project and network level.