The conventional evacuation modelling assumes that that family members are to be together in the face of tsunamis and evacuate as a household or each person evacuate individually. This assumption, however, can cause an inaccurate prediction when taking family gathering behaviour into account, which was reported to have occurred during the 2011 earthquake off the Pacific coast of Tohoku in Japan. This thesis first investigates the family gathering behvaiour during the event using the survey data of evacuation by the survivors of the 2011 Tohoku tsunami. It aims to identify statistically significant travel choices and parameters defining travelers who performed the family gathering during the evacuation. Using the result of this first part, this thesis also suggests a methodology to model that family gathering behvaiour and performs a simulation for a coastal city in Japan using a general activity-based transportation model. It was found as for the behavioral analysis that stay / leave choice, departure time choice, and mode choice shows significant differences between those who performed the family gathering and those who did not whereas concerning the parameters defining those who performed the family gathering initial location, gender, and age have been found significant. The choice frequencies of alternatives developed in the behavioral analysis have been applied into the model formulation taking these findings into account. Depending on whether they performed the family gathering or people's attributes, different choice frequencies have been applied. The simulation indicates that the characteristics of the family gathering trips in comparison with the evacuation trips are explained by the faster speed achieved by the earlier departure time and by the longer travel distance caused by the destination choice being fixed locations such as home and relative’s home rather than nearby buildings. A scientific contribution has been made in terms of evacuation behaviour and evacuation modelling methodology for the family gathering during evacuation. Also, taking this particular behaviour into account for evacuation modelling leads to better prediction of evacuation trips, which in turn helps develop enhanced evacuation planning.