Death anxiety is presumed to be positively associated with group identification; however, recent evidence of a null correlation between the two constructs raises questions regarding this assumption. In contrast to the traditional linear perspective, we proposed and tested a J-shaped curvilinear association that only death anxiety beyond a certain threshold predicts group identification. Using two-wave longitudinal data from the UK, study 1 ( = 1,402) revealed that only after reaching a moderate-to-high level could death anxiety measured during the COVID-19 pandemic positively predict later identification with the community, one's country, and all humanity. Furthermore, using World Values Survey data, study 2 ( = 56,871) found that death-related anxiety (i.e., worry about a terrorist attack) was only positively associated with perceived closeness to one's village, county, and country after reaching a moderate-to-high level. Our findings provide a novel insight into the process of managing terror and the replication failure of the mortality salience effect.