Background: Mentalisation (also known as theory of mind) difficulties have been reliably demonstrated across different subtypes of adult acquired brain injury (ABI), and the role of such impairments in negative psychological and interpersonal outcomes for survivors and their significant others has been increasingly highlighted. Aims & Methodology: This study aimed to characterise the most salient aspects of mentalising performance in a large ABI sample, relative to matched controls. The participants were 88 (64 male, 24 female) persons with acquired brain injuries (TBI; CVA; other subtypes) participating in community neuro-rehab services (mean age 45.2 years, SD 10.7; mean time since injury 6.69 years; range 1.5 – 31.3 years).) and 50 (34 male, and 16 female) healthy participants (mean age 45.3 years, SD 13.9). The main measure of mentalising operationalised in this study was the Recognition of Faux Pas Test (Stone et al., 2003), a story vignette task completed by patients and controls. Results & Conclusions: Overall, the patient group made significantly more errors in detecting the presence of a faux pas than the matched control group (t (132)=2.24, p<.05, Cohen's d = 0.4), reflective of 1st order mentalising difficulties in the ABI group. However the patients did not make more errors than controls in explaining the reason for the faux pas (p=.75). Patterns in errors made by the patient group are explored, and implications for rehabilitation are discussed.