The aim of this study was to test the validity of the CHR state by focusing on the course of psychosis spectrum symptoms, depression and global functioning in converters and non-converters. A total of 188 CHR-positive subjects (60.2% men) aged between 13 and 35years (mean=20.5) at study outset were assessed five times (t0-t4) over a total observation period of 36months. Conversion to manifest psychosis was defined according to ICD-10 criteria for schizophrenia (F20) or brief psychotic disorder (F23). Measures of positive and negative symptoms were assessed with the Structured Interview for Prodromal Syndromes (SIPS), depression with the Calgary Depression Scale (CDS), and global functioning with the Global Assessment of Functioning Scale (GAF). Converters scored higher over time on all SIPS scales apart from grandiosity (Cohen's d: 0.5-0.7; all p<0.001), higher on the CDS (d=0.43, p=0.001) and lower on the GAF (d=0.69, p<0.001) than did non-converters. Positive and negative symptoms as well as depression were most severe at study outset (t0) and then declined sharply following a linear function over the three-year observation period (t1-t4) across groups (all linear contrasts p<0.001). In conclusion, converters showed significantly more psychopathological symptoms and poorer functioning before crossing the diagnostic threshold for manifest psychosis. CHR-subjects who convert to manifest psychosis during follow-up appear to be recovering from illness rather than becoming ill. Major issues involve the poor discrimination of CHR state and psychosis as well as the dichotomous definition of both at-risk and disease states. Further examination in other CHR-samples is warranted.