Abstract: By using floating car technology on urban expressway ,time series speed data were obtained .The time series speed data were processed and then used to analyze and quantify the formation and dissipation of the recurred con -gestions on the bottleneck sections of Beijing expressways .First ,Wavelet analysis method was applied to reducing the noise in the time-series data .The traffic bottleneck condition can be divided into four states ,namely ,steady-state opera-tion ,formation of congestion ,congested state ,and dissipation of congestion according to the characteristics of speed vari-ation .Each traffic bottleneck state can be determined by applying the speed variation amount as the evaluation index .Sta-tistical method was used to determine the threshold from one state changing to another state .The index and threshold can be used to induct the traffic congestion formation time ,dissipation time and duration of congestion .The accuracy rate of the method for identifying traffic congestion is more than 80 percent .The method can provide significant practical benefit for congestion prediction and traffic management .