In new-tier cities with rapid urbanization, the reorganization of urban spatial functions and the development of road networks have brought novel challenges to traffic congestion control. Urban land use patterns have a significant correlation with urban traffic congestion. However, whether and how land use patterns of cities close to the roads affect road congestion is less to be discussed. This article investigated the relationship between land use patterns close to the urban trunk road network and traffic congestion in new tier cities Xi’an, China. We adopted the DBSCAN algorithm to cluster POIs and use the mixed POI clusters to label the socio-economic functions of roads. We found the spatial heterogeneity of POIs on the trunk road network and identified the impacts of the scales and types of POI on road congestion based on the empirical analysis. Compared to the POIs as origin and destination of the trips, the POIs as stopover points of the trips cause significantly more road congestion. The POIs with bidirectional flows at entrances/exits are more likely to cause road congestion than the POIs with unidirectional flows. Moreover, the POIs with flexible traffic flows increase road congestion, while the POIs with predicted traffic flows have no statistically significant correlation with road congestion. The results help urban planners to plan the scale, type, and location of POIs close to roads and to optimize the socio-economic functions of roads and alleviate road congestion.