The aim of this naturalistic study was to evaluate the potential influence of the duration of untreated illness (DUI)--defined as the time elapsed between the occurrence of the first mood episode and the first adequate pharmacological treatment with mood stabilizers--on the clinical course of bipolar disorder (BD). Three hundred and twenty outpatients (n = 320) with a DSM-IV diagnosis of BD--either Type I or Type II--were interviewed; their clinical features were collected and they were naturalistically followed-up for 5 years. At the end of the follow-up observation, the sample was subdivided into two groups: one group with a DUI < or =2 years (n = 65) and another group with a DUI >2 years (n = 255). The main demographic and clinical variables were analyzed and compared between the two subgroups of patients using chi-square tests for dichotomous variables or Mann-Whitney U tests for continuous variables. Patients with a longer DUI showed a higher frequency of suicide attempts (Z = -2.11, P = 0.035), a higher number of suicide attempters (chi(2) = 4.13, df = 1, P = 0.04), and a longer duration of illness (Z = -6.79, P < 0.0001) when compared to patients with a shorter DUI. Moreover, patients with a longer DUI had a depressive first episode more frequently than patients with a shorter DUI (chi(2) = 11.28, df = 2, P = 0.004). A further analysis performed dividing the total sample into two subgroups on the basis of a DUI of 6 years (corresponding to the median value of the DUI in the study sample) confirmed prior findings. Results indicate a potential association between a longer DUI and a worse outcome in BD, particularly in terms of suicidality, and confirm the clinical relevance of early diagnosis and pharmacological intervention with mood stabilizers in BD.