Cognitive reserve (CR) has been associated with the development and prognosis of psychosis. Different proxies have been used to estimate CR among individuals. A composite score of these proxies could elucidate the role of CR at illness onset on the variability of clinical and neurocognitive outcomes. Premorbid intelligence quotient (IQ), years of education and premorbid adjustment were explored as proxies of CR in a large sample (N = 424) of first-episode psychosis (FEP) non-affective patients. Clusters of patients were identified and compared based on premorbid, clinical and neurocognitive variables at baseline. Additionally, the clusters were compared at 3-year (N = 362) and 10-year (N = 150) follow-ups. The FEP patients were grouped into five CR clusters: C1 (low premorbid IQ, low education and poor premorbid) 14%; C2 (low premorbid IQ, low education and good premorbid adjustment) 29%; C3 (normal premorbid IQ, low education and poor premorbid adjustment) 17%; C4 (normal premorbid IQ, medium education and good premorbid adjustment) 25%; and C5 (normal premorbid IQ, higher education and good premorbid adjustment) 15%. In general, positive and negative symptoms were more severe in the FEP patients with the lowest CR at baseline and follow-up assessments, while those with high CR presented and maintained higher levels of cognitive functioning. CR could be considered a key factor at illness onset and a moderator of outcomes in FEP patients. A high CR could function as a protective factor against cognitive impairment and severe symptomatology. Clinical interventions focused on increasing CR and documenting long-term benefits are interesting and desirable.