Objective This study assessed lay people and clinicians’ recognition of an at risk mental state (ARMS) for psychosis, their intentions to recommend help, and the equivalence of written and videotaped vignettes when detecting the problem. Methods In an Australian online survey, 52 lay people and 32 psychologists with provisional or full registration were randomly assigned to either a videotaped or written vignette of someone with an ARMS. Measures assessed detection and labelling of the mental health problem, and lay people's intentions to recommend help. Results Data were analysed with Chi‐Square statistics, Fisher's Exact tests, and Multinomial Logistic Regression. Lay people frequently detected that a mental health problem existed but labelled it incorrectly. All clinicians detected that a mental health problem existed and most labelled it correctly. Lay people's detection that a mental health problem existed was not associated with vignette type but videotaped vignettes produced significantly more correct labelling. Clinicians had poorer labelling when the vignette was videotaped. Correct labelling was associated with intentions to recommend help to a doctor, psychiatrist, and psychologist/counsellor but not with other help sources or with “no help.” Conclusions Results indicated that if lay people received further education about ARMS, they may be more likely to recommend help to certain mental health professionals. They further highlight the need to use multiple vignette methods in mental health literacy research and the importance of simulated learning about ARMS in professional training environments. Replication of these results in larger samples is required.