The duration of untreated illness (DUI), meaning the latency to the pharmacological treatment, has been increasingly investigated in the last decade as a predictor of outcome across different psychiatric conditions, particularly in psychotic disorders. DUI is essentially computed by subtracting the age of onset of a specific disorder from the age at which the first adequate pharmacological treatment is administered. Assessment of the latency to treatment represents one of the first steps in planning early interventions. This review examines the role of the DUI in psychotic and affective disorders, focusing on neuropathological, epidemiologic, clinical and prognostic factors related to a longer latency to treatment. Through a Medline and Cochrane Library search, relevant studies up to June 2011 and other pertinent articles including meta-analyses, randomized controlled trials, naturalistic studies and clinical reviews were identified. Converging evidence indicates that a prolonged DUI negatively influences the outcome of first-episode psychosis and schizophrenia in different ways, and increasing data point toward a similar conclusion in affective disorders. Even though methodological limitations related to investigation of the DUI need to be considered, research and interventions aimed to reduce latency to treatments are object of increasing implementation worldwide. The assessment of the DUI represents one of the most important parameters to consider in this perspective, in order to quantify different latency to treatment in specific disorders and to plan related, targeted interventions.