This paper proposed a methodology to estimate rear-end crash potential of the merging vehicles traveling in the merge lane, on the basis of the traffic data extracting from the available videotapes. First, we developed a binary logit model to identify drivers’ merging behavior in the work zone merging area. Subsequently, the occurrence potential of rear-end crash based on time-to-collision was computed between the merging vehicle and its neighboring vehicles. The overall crash potential of the merging vehicle was finally determined. It was found that the crash potential decreases with the remaining distance to work zone. Moreover, there will be a rear-end crash potential of 4.0% if the merging vehicle fails to complete merging at the end of work zone merging area. If the merging vehicle takes an early merge, there will be a lower rear-end crash potential (1.2%). These findings suggest that we should encourage merging vehicles to take early merges for improving the traffic safety in the work zone merging areas.