Psychiatric illnesses such as bipolar disorder, schizophrenia and schizoaffective disorder are severe, disabling disorders associated with decreased Quality Of Life (QOL) and functioning. Stigmatization, co-morbidities, adverse effects of medications and chronic symptoms due to treatment resistance are factors responsible for this association. In this study, we aim to characterize patients with good and poor outcomes according to QOL and functioning scores. Using cluster analysis, we sought to identify longitudinal trajectories and investigate whether levels of QOL and functioning are associated with polygenic risk scores. Determining clusters of patients at higher risk of poorer outcomes is critical to provide early and effective interventions. Longitudinal data was used from the Clinical Research Group 241 and PsyCourse studies in Germany. Participants were phenotyped using a comprehensive battery which included data on socio-demographics, history of illness, symptomatology, QOL and functioning. Data was collected at four equidistant time points over 18 months. The Infinium Psycharray from Illumina was used to genotype patients. The QOL and functioning domain was represented by 28 longitudinally measures items (WHOQOL-BREF; GAF; work status). Dimension reduction of the 28 items by means of Factor Analysis for Mixed Data (FAMD) was applied. This allowed for the computation of abstract data dimensions which were used for calculation of longitudinal trajectories. These trajectories are a representation of the overall status of patients and both the overall level as well as the longitudinal change of this status were used as inputs for k-mean clustering of longitudinal data. This, in turn, resulted in the identification of three distinct subpopulations of patients. A multinomial regression of cluster membership on schizophrenia Polygenic Risk Scores (PRS) (11 p-value thresholds) was performed. Individual trajectories across time on nine dimensions were used for cluster analysis. Together, these dimensions explained 27.13% of variance in the data. These dimensions were mainly driven by QOL items. In a sample of 254 patients, three clusters were observed; cluster A (63%) consisted of participants with the highest average scores for functioning, cluster B (33%) and cluster C (4%) consisting of participants with lower functioning over the course of the longitudinal study. A significantly higher proportion of individuals in cluster B were hospitalized at baseline in comparison to cluster A. Furthermore, individuals in cluster A had a significantly higher GAF score at baseline than those individuals in cluster B. No significant differences were seen for age, center, diagnosis, sex, work status, or duration of illness between the clusters. A nominally significant effect of SZ-PRS on cluster membership was observed from the threshold 0.2 on, however this did not survive FDR correction. In terms of cluster membership, the highest polygenic risk loading was observed for cluster B and the lowest for cluster C. Phenotypic data provides insight to target sufferers of severe mental illness with worse outcomes. Further work is needed to improve methods to deal with missing data and increase sample size. Associations with other biological data will be explored (e.g. microRNA; methylation data; proteomics; imaging data).