The current study investigated the relationship between therapist and client attachment styles and perceptions of therapeutic alliance in a sample of inpatients with psychosis. A review of the literature found that attachment theory can provide understanding regarding the behaviour of clients with psychosis in therapeutic relationships, especially in inpatient settings. Working alliance can be measured as a representation of the therapeutic relationship and both client and therapist attachment styles contribute to the working alliance and can be measured in various ways. Research has suggested that for clients with psychosis diagnoses dissimilarity between client and therapist attachment styles indicated a better client rated working alliance. Given the low number of studies in this area and the relevance of attachment theory in working with clients with a psychosis diagnosis, this study aimed to investigate the relationship between therapist and client attachment styles and client perceptions of working alliance in therapy when clients had a diagnosis of psychosis. The present study hypothesised that clients’ with psychoses and therapists who were dissimilar in terms of attachment styles would predict better client perceptions of working alliance. The current study focused on clients with a diagnosis of psychosis because distress, trauma and interpersonal difficulties are prominent features of psychosis, which would influence individuals’ attachment behaviours and outline a role for attachment theory in understanding difficulties and tailoring treatments for this client group. The current research decided to focus on clients with psychosis in an inpatient setting to extend previous research which has mainly included clients in a community setting. Attachment theory can play an important role in considering the inpatient environment and how it can influence attachment behaviours. The present study was a cross sectional, within subjects design. Therapists and clients rated their attachment styles and working alliance. Most therapists rated their attachment style and working alliance with more than one client, therefore client data was nested within specific therapists. Using the Relationship Questionnaire (RQ) and Working Alliance Inventory (WAI), 46 pairs of therapists and clients recruited from independent inpatient psychiatric hospitals were assessed. A regression analysis examined the relationship between client and therapist attachment styles and client rated working alliance. The nested structure of the data and therapist rated working alliance was controlled for in the analysis. The present study did not find a significant relationship between the difference between therapist and client attachment styles and client rated working alliance. The present study hypothesised that clients’ with psychoses and therapists who were dissimilar in terms of attachment styles would predict better client perceptions of working alliance. Therefore, the current study did not confirm this hypothesis. It is considered that perhaps clients had not had enough time to build up the therapeutic relationship with therapists to allow the difference in attachment styles to lead to a better working alliance, meaning that more time may be needed in the initial stage of therapy. Further research is recommended to tease apart the attachment style interaction, including considering other variables and measuring working alliance over time in longitudinal studies.