Background: Baseline functioning has been found to be a strong predictor of transition to psychosis in ultra high risk populations. However, the time course of functioning may enhance prediction. We investigated whether there were different patterns of functioning over time and whether particular temporal patterns were related to baseline characteristics and psychosis outcome. Method: Functional data was assessed at baseline and after 3 to 6. year follow-up in an ultra high risk sample (n = 158; 92 female, mean age = 19.28 (SD = 3.33), range = 14-29). Using the median score of the GAF and the QLS scale, a 'High' and 'Low' group (comprising of subjects functioning above or below median at both baseline and follow-up) and a 'Deterioration' group and 'Improving' group were created. Results: Chi-square analyses showed that the Low and Deteriorating functioning groups were the most likely to develop first-episode psychosis (FEP). Importantly, UHR individuals with deteriorating functioning were at higher risk of transition than those whose functioning was low at baseline but improved over time (GAF: X2=5.10, df=1, p=.02; QLS: X2=9.13, df=1, p=.003). Binary logistic regression analyses showed that a decline in functioning was more strongly associated with FEP (GAF: p=<.0001; QLS: p<.0001) than the level of baseline functioning (GAF: p=.005; QLS: p=.09). The deteriorating group could not be distinguished from the High group in terms of baseline symptomatology. Discussion: With the addition of the 'low functioning' criterion to the UHR criteria, we may miss out on some true positive cases. Limiting our attention to baseline poor functioning may therefore distort the picture in terms of risk for psychosis.