<p>The issue of urban traffic congestion is a significant obstacle in the continuous process of urbanization. Although there have been extensive studies conducted to address the issue of traffic congestion, challenges and limitations still exist. For example, it is hard to provide good interpretations and reveal spatio-temporal correlations, and there are limitations in road-level congestion propagation. Spatio-temporal co-location pattern mining has gained significant popularity in the field of spatio-temporal data mining in recent years, which allows the discovery of meaningful patterns with spatio-temporal correlation. Therefore, we first propose the concepts of regional congestion propagation patterns (RCPPs) based on spatio-temporal co-location patterns to discover congestion propagation patterns with spatio-temporal correlations. Second, we propose to grid the urban road network and consider grid regions as features, which can analyze congestion propagation at the regional level. Third, we propose a mining framework and two algorithms for mining prevalent RCPPs. The experiments with the road network in Beijing and real road conditions data indicate that our research is meaningful and efficient.</p>