Abstract Objective. To assess whether a long duration of untreated psychosis (DUP) before first admission predicts poor clinical and social outcome, and whether this association, if any, is confounded by premorbid and clinical characteristics. Method. A population-based sample of first-admitted subjects with psychosis ( n=65) was assessed at six monthly intervals over a two year follow-up using multiple sources of information. Results. Most subjects (87%) with a life-chart ‘continuous’ course of psychotic symptoms had a history of a ‘long’ delay between onset of psychotic symptoms and first admission (≥3 months, median split), compared with 55% of subjects with a course of ‘neither episodic nor continuous’, 42% of subjects with an ‘episodic’ course, and 33% of subjects with ‘no psychotic symptoms’ during the follow-up period (RR=9; 95%CI 1.5–54.8, P=0.02). The strength of association between DUP and continuous course of psychosis was strongly reduced (63%) after adjustment for premorbid functioning, and to a lesser extent for the severity of illness and for the intensity of negative symptoms at first admission. Conclusions. The association between DUP and poor outcome may be spurious, confounded by the fact that poor premorbid functioning is independently associated with both DUP and poor outcome, with no direct causal link between these two latter variables. DUP may also be on the causal pathway between poor premorbid functioning and poor outcome, poor adjustment delaying access to care, and subsequently increasing the risk of presenting with a non-remitting course of illness. The links between premorbid functioning, DUP and outcome have to be further explored to clarify the directions of the associations between these variables.