Abstract Background: Neurocognitive impairment is a core feature of schizophrenia and is observed in the majority of patients. Substance use is common among patients with psychosis and linked to poorer outcomes. As neurocognition is one of the strongest long-term predictors of functional outcome and recovery poorer outcomes would be expected for substance using psychosis patients. However, literature shows mixed results for the effect of substance use on neurocognitive outcomes and establishing causality has proved challenging. The aim of our study was to examine long-term influence of substance abuse with regards to neurocognition in a representative sample of first-episode psychosis (FEP) patients over 10 years, hypothesizing that patients who continue using will have poorer function and higher symptom loads than those who do not use substances, that patients with substance use will follow a different trajectory than nonusers and will not have inferior scores on neurocognitive testing compared to nonusers. Methods: Three hundred and one FEP patients, 188 without substance use, and 113 with use from the TIPS study were assessed for baseline characteristics, symptom levels, function, and neurocognition at 1, 2, 5, and 10 years. Neurocognition was analyzed through linear mixed-effects modeling. Results: Substance users were predominantly male (77.9% vs 46.8%; P < .000), younger at presentation than nonusers (24 vs 30 y; P < .000), and had fewer years of education (11.4 vs 12.2 y; P < .005) as well as having better social function premorbidly (P < .014) and poorer function on the GAF at 10-year follow-up (P < .05) with less abusers fulfilled criteria for recovery (P < .05) at 10 years. Symptom scores on PANSS were higher in substance users at 10-year follow-up with regards to the positive component (P < .017), cognitive component (P < .013), and the excitatory component (P < .001). Linear mixed-effect analyses of neurocognitive measures only showed differences with regards to motor speed (P < .012) with time and gender as significant covariants. Conclusion: Our sample of FEP patients followed up over 10 years show that despite substance users having higher symptom loads and poorer function there is little difference in neurocognitive function between groups. This paradoxical finding in substance users presents us with several possible explanations. The difference could be attributed to social skills required to gain access to illegal substances; however, this is not consistent with this patient group showing poorer outcomes in daily function and symptoms. Alternatively, substance users could represent a separate group with different vulnerabilities and trajectories. This finding is consistent with other studies, but few have examined this important domain directly and there is a need for further research in to this area.