Background Recently, school mobility was identified as a risk factor for psychotic symptoms in early adolescence. The extent to which this risk continues into late adolescence and the trajectories via which this risk manifests remain unexplored. Methods Psychotic symptoms in 4,720 adolescents aged 18 were ascertained by trained psychologists using the Psychosis‐Like Symptoms Interview. Mothers reported on sociodemographic factors (i.e., family adversity, ethnicity and urbanicity) from pregnancy to 4 years; child's involvement in bullying at age 6–7 years; residential mobility at 11 years and school mobility at 11–12 years. Young people reported on their friendships at 8 years, and antisocial behaviour and cannabis use at 15 years. Results School mobility across childhood significantly predicted psychotic symptoms at 18 years (adjusted odds ratio = 2.15; 95% confidence intervals = 1.06, 4.40). Within path analysis, school mobility (β = .183, p = .035), involvement in bullying (β = .133, p = .013), antisocial behaviour (β = .052, p = .004), cannabis use (β = .254, p = .020) and female sex (β = .420, p < .001) significantly predicted psychotic symptoms. Residential mobility (β = .375, p < .001), involvement in bullying (β = .120, p = .022) and poor friendships (β = .038, p = .014) significantly predicted school mobility. Residential mobility indirectly increased the risk of psychotic symptoms via school mobility (β = .069, p = .041). Conclusions Children who move schools often are more likely to have experienced peer problems. School mobility, in turn, appears to be a robust marker for psychotic symptoms in late adolescence. Clinicians and teachers should consider school mobility as an important risk indicator for both peer problems and psychopathology.