The invention discloses an urban region road network running state characteristic information extraction method. The method comprises steps that, step 1, the road network data is acquired; step 2, a road network data input matrix is constructed; step 3, adaptive neighborhood selection is carried out; step 4, characteristic information extraction is carried out; and step 5, visual representation of the road network running state characteristic information is carried out. The method is suitable for urban region road network characteristic information extraction and road network running state representation, on the basis of road network flow, speed and density data, an adaptive neighborhood selection manifold learning method is utilized to extract the road network running characteristic information, the road network running characteristic information has macroscopic property, accuracy, sensitivity and practicality, a macroscopic road network running state can be timely represented, and analysis and decision bases are provided for traffic managers.