AbstractMost initial implementations of pavement management systems face limited or unavailable local prediction of treatment effectiveness. This means decision makers either are forced to borrow benchmark values from elsewhere or should wait at least 5 years until time series data are collected, enabling them to measure effectiveness locally. This research proposes an initial approach to help agencies predict the effectiveness of maintenance interventions with limited data while data collection is under way and traditional time-series analysis is not yet available. The proposed method is applied to the Montreal island road network, which is constantly criticized as one of Canadian cities’ worst-condition road networks. This study uses Markov transition probability matrices by functional classification and type of pavement structure. The provided matrix shows that preventive maintenance on arterial roads achieved a service life extension of almost 2 years, while a similar approach for local roads gained less. Furthermore, although 40% of rigid segments were arterial roads, a similarity was observed between the rigid road matrix and rigid local matrix, which shows planning based on type rather than the road’s function. The outputs of the developed model in this study suggest that it is possible to create roads with a better condition level by estimating intervention effectiveness and optimal timing. This study contributes to improving asset management decision-making tools for roads and is extensible to other types of infrastructure, especially those under municipalities and railway systems.