This paper concerns the modeling of time headway distributions on the A6 Motorway in France. The aim of the work is to compare these distributions in different contexts such as type of lane, traffic flow, period of the day and change of cross-profile. In order to interpret and to distinguish the different contexts and behaviors, the composite time headway distribution Generalized Queuing Model and the best simple model, the Log-normal Model, are calibrated. In addition, the statistical analysis of a series of events is applied to understand the time headway process. The distribution of instantaneous speeds enriches such interpretations in several cases. The results show a significant goodness-of-fit of the composite time headway distributions with most samples. The Log-normal Model provides also good adjustments for many samples particularly on the middle and fast lanes. Consequently, the sensitivity of the parameters of these models is yielded according to the contexts.
