Abstract: Identification of traffic breakdown is critical for evaluating traffic flow states and making decision on traf-fic operation plan.In order to detect the differences of traffic conditions among the lanes in detail,a lane-based method is proposed,which is able to identify breakdown accurately and in an intuitive way.The occurring time of traffic breakdown is determined by enlarging the"turning point"of speed fluctuations at each time interval using speed transform cumulative curve.Compared to the traditional section-based and threshold-based methods,the proposed lane-based method is more accurate and objective which can not only differentiate the semi-congested states,where only some of lanes are congested, but also show the longitudinal and transverse propagation characters of traffic congestion in depth.The off-ramp at the Jungong Expressway Road in Shanghai is used as a case study and the results show that the breakdown actually occurs on the outermost lane firstly and then spread to innermost lane after an average of 4 minutes.When the congestion propa-gates to the upstream,the inner lanes tend to be more congested than the outer lanes.