According to research applying a social constructionist perspective, the sense of self is not lost in people with Alzheimer’s disease (AD). It is, however, greatly influenced by the symptoms and by how they are treated by other people. Without support, it is difficult to preserve a positive sense of self, when living with progressing cognitive impairments. The stigma associated with cognitive impairment also threatens their sense of self. Harré’s social constructionist theories of self and positioning have been used to study how people with AD express their sense of self. As there is a need to expand the previous research by involving additional participants and research contexts, the aim of the present thesis was to describe, in accordance with Harré’s theories of self and positioning, how people with AD expressed their sense of self in personal interviews and in support groups with other people with AD. The research consists of four substudies (I–IV), and has a qualitative, descriptive, and theory-testing approach. Thirteen people with mild and moderate AD were included, 11 of whom had the early onset form of the disease. Two support groups were formed, led by facilitators who supported the communication and the participants’ expressions of self. Each group met 10 times during an eight-month period. Topics were not predetermined, and introduced by both facilitators and participants. Semistructured interviews were conducted before the groups started and after they ended. The interviews and support group conversations were audio-recoded and analysed with qualitative content analysis, guided by Harré’s theories. In substudy I, the initial interviews were deductively analysed. The findings showed that Self 1 (the sense of being a singular, embodied person) was expressed by the participants without difficulties. Self 2 (the perception of one’s personal attributes and life history) was expressed as feeling mainly the same person. While some abilities had been lost, other had been developed. Self 3 (the socially constructed self) was described as mostly supported, but sometimes threatened in interactions with other people (I). In substudy II, support group conversations were analysed abductively with respect to expressions of Self 2. It was found that participants expressed Self 2 in terms of agency and communion, and a lack of agency and communion (II).In substudy III, a secondary analysis of the data from substudy II was performed inductively with the aim of describing how Self 3 was constructed in the interaction of the support group. Five first-order positions, generating lively interaction, were described: the project manager, the storyteller, the moral agent, the person burdened with AD, and the coping person (III). In substudy IV, all the collected data were reanalysed inductively, focusing on how participants expressed the experience of being research participants. Three themes were constructed: contributing to an important cause, gaining from participating, and experiencing risks and drawbacks (IV). In conclusion, it was found that participants constructed positive social selves through the support from each other, the facilitator, and researchers in the support group (III), and as research participants (IV). Agency and communion were central to Self 2, and decreased with the progression of AD (II). In spite of change, participants perceived themselves as basically the same people, with a potential to learn and develop as persons (I).