Background/Objectives: Describing and predicting the developmental course of disease in patients with severe mental illness is a central theme in psychiatric inquiries. Latent growth modelling approaches, such as latent class growth analysis, have been increasingly recognised for their usefulness in identifying homogeneous subpopulations within the larger heterogeneous population as well as meaningful groups or classes of individuals. This study focuses on latent class analysis as a means to represent the various courses of disease in patients with schizophrenia and schizoaffective disorders. We examine two research questions: (1) Is it possible to identify latent subgroups of patients characterized by different shapes of trajectories?, and (2) Is it possible to predict the class membership by patients’ sociodemographic and clinical characteristics? Methods: In a prospective longitudinal study with 5 points of measurement over a period of two years, the course of disease of 374 adults living in Southern Germany with a diagnosis of schizophrenia or schizoaffective disorders was assessed. Interviews were conducted face-to-face by clinical psychologists trained in using the study instruments for assessing study participants’ psychopathology (Positive and Negative Syndrome Scale), global functioning (Global Assessment of Functioning Scale), quality of life (Berliner Life Quality Profile), and cognitive functionality (Digital Symbol Coding Subtest from the Wechsler Adult Intelligence Scale). Latent Class mixture models were used to identify patterns of change. These models aim to uncover unobserved heterogeneity in a population and to find substantively meaningful groups of people that are (almost) equal in their responses to measured variables or growth trajectories. Results: The latent growth modelling of change patterns resulted in three types of clinically relevant and significant classes of change processes. Those patterns were related to specific patient variables at study entrance and were also associated with differenzial indicators of life circumstances. In addition, the results confirmed the mediating role of cognition; cognitive deficits are a core feature of schizophrenia and the results suggested that they can significantly predict clinical and social outcome in schizophrenia patients. Discussion/Conclusions: The findings suggest utility in using the latent class analysis in order to better understand predictors of course of disease. Results may inform the development of more effective prevention and intervention efforts targeting patients on different disease trajectories. Thus, the modelling of change patterns in people with schizophrenia can be helpful in identifying negative and positive treatment developments. Further applications of these methods as well as clinical relevance are discussed. Funding: The ELAN study was funded as an investigator-initiated research project by a grant from AstraZeneca Deutschland to the University of Tubingen (grant no. 229/2004V). Keywords: Patient outcome assessment, course of disease, longitudinal study.