Research on the complexity of traffic flow evolution is helpful to deeply understand the evolution rule of traffic flow system, which can provide the theoretical foundation for forecasting and controlling traffic flow. Multi-scale entropy (MSE) method is widely used in the analyses of time series of physiology and traffic of computer networks. Considering the similarity between the vehicle arrival in traffic flow system and the packet arrival in computer network, the complexity of the time headway in braking light model is analyzed to show the complexity of traffic flow evolution by using the MSE method. The analysis results show that the complexity of the time headway decreases with the increase of the time scale, which reflects that it is difficulty to predict the traffic flow in a shorttime. In addition, the difference in the complexity of the time headway between the phases of the free flow and synchronized flow is small when the time scale is small. However, with the increase of the time scale, the MSE of the time headway decreases rapidly for free flow, but rather slowly for synchronized flow. Such a difference can be used as a very important reference to distinguish the synchronized flow and the free flow. Research results in this paper can provide new ideas and methods for investigating the complexity of traffic flow evolution.