This paper presents two empirical trajectory data sets focusing on the merging behaviour on a motorway, both in the Netherlands and in France. A careful review of the literature shows that the main theories explaining this behaviour rely on the hypothesis of gap acceptance, i.e. the fact that each driver has a certain threshold value depending on among other things the distance to the end of the acceleration lane, and when the offered gap is larger than this threshold the driver decides to merge. We conducted a detailed comparative analysis of the two data sets examining the main variables identified in our conceptual model of merging behaviour. The contribution of this paper is that the analysis does not only focus on the accepted gaps, but it also takes into account the rejected gaps. The comparison of our observations with the critical gap formula in literature showed that this formula does not take into account the strong probability of rejecting a gap, even larger than the gap finally accepted. Moreover, we created a logistic regression model that predicts the acceptance or rejection of a given gap, depending on the gap value and the speed difference between the merging vehicle and the putative follower. We have shown that two other factors impact the probability of rejecting or accepting a given gap, but these are significant for just one of the data sets: the distance to the end of the acceleration lane and the speed difference between the putative follower and the putative leader. This shows the impact of the local situation on the merging behaviour (e.g. traffic composition, road geometry, and traffic conditions).