Background: The under-representation of women and other minority group members in STEM (Science, Technology, Engineering, and Mathematics) academia is a problem internationally and is attributed in part to hostile workplace cultures. We draw on the social identity perspective to examine the dynamic inter and intragroup processes entailed in these experiences. In this paper, we report a reflexive thematic analysis of 219 responses to a free-text question on bullying and harassment embedded in a national survey of 40 STEM departments from across the United Kingdom. Most were women (53%) at an early, pre-lectureship career stage. Main findings: Our analysis shows who is the perpetrator and who is the victim is not arbitrary, and neither is the form that it takes; majority group members draw on discourses that warrant the exclusion of minority group members (e.g., women are not smart; incompatibility with religious identity). In this 'othering', minority group members learn that one is not regarded as a bona fide STEM academic ingroup member and accordingly are constrained in being able to claim and act on that identity. Thus, it is not just the acts themselves that are problematic, but the ways in which being denied a shared STEM academic identity is consequential for a range of putative benefits and leads to a range of strategies that all confer costs. Conclusions/potential implications: The solution must rest with senior STEM academics and with institutions. First, we need to challenge discourses and practices that narrowly define the boundaries and content of STEM academic identity. Second, all members of a community need to perceive an alignment between the purported values of an organisation for diversity, inclusion, and respect and how that organisation responds when those principles are violated. Formal processes of remedy need to recognise the dynamics entailed in status differences and remove the onus of complaint from isolated, low status individuals. In addition, there is a need to recognise the ways in which perpetrators are embedded in networks of support both within and without the university; and the importance, therefore, of widening the scope of evidence gathering and intervention.