Insomnia has been shown to contribute to the development of psychotic experiences, predominantly via increasing negative affect. However, the role of insomnia in the persistence of psychotic experiences is yet to be investigated in a clinical population. Furthermore, other plausible influences, such as psychotic experiences contributing to insomnia, remain to be evaluated. This study tests the role of insomnia as a predictor of persistence of psychotic experiences versus other potential causal routes. Twenty-nine patients aged 18–30 with non-affective psychosis completed three assessments over three months of their insomnia, negative affect, and psychotic experiences. Mixed effect models allowed comparisons between hypothesis-based models (comprising insomnia as predictor, negative affect as mediator, and psychotic experiences as outcome) and oppositional models, where relationships were reversed. The results supported the hypothesised mediation model above models where negative affect was primary. Insomnia was also found to be a stronger predictor of later hallucinations than vice versa, although a bidirectional relationship was indicated between insomnia and paranoia. In conclusion, insomnia predicts persistence of psychotic experiences over time to the same or greater extent than psychotic experiences contribute to insomnia. This supports insomnia as a potential intervention target in psychosis.