Biases in cognition such as Jumping to Conclusions (JTC) and Verbal Self-Monitoring (VSM) are thought to underlie the formation of psychotic symptoms. This prospective study in people with an At Risk Mental State (ARMS) for psychosis examined how these cognitive biases changed over time, and predicted clinical and functional outcomes. Twenty-three participants were assessed at clinical presentation and a mean of 31 months later. Performance on a JTC and VSM tasks were measured at both time points. Relationships to symptom severity, level of function and the incidence of psychotic disorder were then examined. The levels of symptoms, function and VSM all improved over time, while JTC was stable. Five participants (22%) developed a psychotic disorder during the follow-up period, but the risk of transition was not related to performance on either task at baseline, or to longitudinal changes in task performance. JTC performance correlated with symptom severity at baseline and follow-up. Similarly, performance on the two tasks was not related to the level of functioning at follow-up. Thus, while the ARMS is associated with both VSM and JTC biases, neither predict the onset of psychosis or the overall functional outcome.