In spite of the harsh conditions that refugees of conflicts experience for many years in exile in developing countries, there is evidence showing that refugees of conflict help each other. This study aimed to explore one possible mechanism underlying such support and sought to answer three main research questions: Do refugees share an emergent identity that facilitates support among them (similar to people affected by disasters)? Does this identity-based support have an impact on their health? If so, does this positive impact help to mitigate the negative effect of exile stressors on refugees' health? We carried out two questionnaire surveys among Syrian refugees, first in Turkey (n = 234) and then in Jordan (n = 156). The data were analysed using path analysis to test hypotheses and build a theoretical model. We found evidence suggesting a process of shared social identity-based support among the refugees. We found that the general health of refugees to be predicted mainly by stress, but we also found that collective efficacy has a positive association with health, which suggests a buffering effect. These results shed light on the process of social support among refugees of war and suggest the role of shared identity, which can have a limited buffering effect on the health of the refugees, though not enough to fully mitigate the negative effect of secondary stressors. However, we suggest that such a process can be utilised as base for interventions that approach refugees of war as a group (i.e. at community rather than individual level).