Up to 80% of first-episode psychosis patients suffer a relapse within five years of the remission. Relapse should be an important focus of prevention given the potential harm to the patient and family. It threatens to disrupt their psychosocial recovery, increases the risk of resistance to treatment and has been associated with greater direct and indirect costs for society.Based on a previous project entitled “Genotype–phenotype and environment. Application to a predictive model in first psychotic episodes” (PEPs Project), the project “Clinical and neurobiological determinants of second episodes of schizophrenia. Longitudinal study of first episode of psychosis” was designed, also known as the 2EPs Project. It aimed to identify and characterize those factors that predict a relapse within the years immediately following a first episode. This project has focused on following the clinical course, with neuropsychological assessments, biological and neuroanatomical measures, genetic adherence and physical health monitoring in order to compare a subgroup of patients with a second episode to another group of patients which remains in remission. The main objective of the present article is to describe the rationale of the 2EPs Project, explaining the measurement approach adopted and providing an overview of the selected clinical and functional measures.2EPs Project is a multicenter, coordinated, naturalistic, longitudinal follow-up study over three years in a Spanish sample of patients in remission after a first-psychotic episode of schizophrenia. It is closely monitoring the clinical course of the cases recruited to compare the subgroup of patients with a second episode to that which remains in remission. The sample is composed of 223 subjects recruited from 15 clinical centres in Spain with experience of the preceding PEPs Study project, albeit 2EPs being an expanded version with new basic groups in biological research. From the total sample recruited, 63 patients presented a relapse (44%).2EPs arose to characterize first episodes in an exhaustive, novel and multimodal way, thus contributing towards the development of a predictive model of relapse. Identifying the characteristics of patients who relapse could improve early detection and intervention.