With the consideration of the influence of management conditions, this paper proposes a pavement performance evolution pattern recognition and management level evaluation method based on cluster analysis to in-depth study the changes in pavement performance. First, the evaluation data of road conditions in a Chinese city was preprocessed to verify the completeness and accuracy. Then, the time sequence of Pavement Condition Index (PCI) and Riding Quality Index (RQI) served as the cluster statistics. An unsupervised learning method was utilized to divide the evolution of pavement performance into five patterns. Last but not least, based on the pavement performance evolution curve, the pavement management and maintenance status is quantitatively evaluated. The results show that there is a correlation between the management level of different road sections and the evolution pattern curve of pavement performance. It provides guidance for pavement maintenance management and also reveals the applicability of the clustering method in data mining in the road administration industry.