Introduction: Several studies suggest the presence of reality-monitoring difficulties, and particularly of an external attribution bias, in schizophrenia (Brébion et al., 2002). This cognitive deficit is exacerbated by emotional material (Allen et al., 2004). However, these results are mostly based on studies evaluating chronic schizophrenic adults. We were therefore wondering if the same cognitive difficulties were present in two groups of adolescents at increased risk for schizophrenia: 22q11DS adolescents and high schizotypal clinical adolescents. Method : We recruited 17 adolescents with 22q11DS, 17 high schizotypal clinical participants and 17 IQ-matched controls. Every participant completed a self-reported questionnaire assessing schizotypal manifestations (SPQ). They also participated in a reality-monitoring task in which they were seeing happy, angry and neutral faces on a computer screen associated with three optional names. Every participant and the examinator took turn in attributing one of the names to each face. In a retrieval phase, they had to recognize the chosen names and identify who had chosen them (self or examinator). Results and discussion : Our results suggest that reality-monitoring performance of 22q11DS adolescents is characterized by an external attribution bias. In addition, we found that participants in the total sample made more emotional than neutral source confusion errors. We also observed clinical correlations between the SPQ and source confusion scores. Taken together, these results suggest the presence of early cognitive markers for psychosis in 22q11DS adolescents. We also discuss the differences obtained between this group and the sample of high schizotypal clinical adolescents.