Identities ascribed to research staff in face-to-face encounters with participants have been raised as key ethical challenge in transnational health research. 'Misattributed' identities that do not just deviate from researchers' self-image, but obscure unequivocal aspects of researcher identity - e.g. that they are researchers - are a case of such ethical problem. Yet, the reasonable expectation of unconcealed identity can conflict with another ethical premise: confidentiality; this poses challenges to staff visiting participants at home. We explore these around a case study of 'follow-up' staff, observed during an ethnographic study of a Kenyan HIV 'trial community', which included participant observation, conversations, and interviews with staff (n = 79) and participants (n = 89). We found that because of the need to maintain confidentiality and because of some suspicions towards researchers, research staff drew upon alternative identities - presenting themselves to non-participants as relatives or friends, rather than as researchers. Several staff experienced this as necessary but uncomfortable. Simultaneously, staff and participants forged close relations in line with their fictional identities, which however also posed challenges because they entailed personal responsibilities that were difficult to live up to, due to limited resources, and the trial's limited duration. Similar challenges may arise in transnational HIV treatment programmes and should be explored further in that context.