The main goal of the present study was to explore the latent structure of extended psychosis phenotypes in a representative sample of adolescents. Moreover, associations with socio-emotional adjustment, academic achievement, and neurocognition performance across the latent profiles were compared. Participants were 1506 students, 667 males (44.3%), derived from random cluster sampling. Various tools were used to measure psychosis risk, subjective well-being, academic performance, and neurocognition. Based on three psychometric indicators of psychosis risk (schizotypal traits, psychotic-like experiences, and bipolar-like experiences), four latent classes were found: non-risk, low-risk, high reality distortion experiences, and high psychosis liability. The high-risk latent groups scored significantly higher on mental health difficulties, and negative affect, and lower on positive affect and well-being, compared to the two non-risk groups. Moreover, these high-risk groups had a significantly higher number of failed academic subjects compared to the non-risk groups. In addition, no statistically significant differences in efficiency performance were found in the neurocognitive domains across the four latent profiles. This study allows us to improve the early identification of adolescents at risk of serious mental disorder in school settings in order to prevent the incidence and burden associated with these kinds of mental health problems.