Objective: There have been few prospective long-term naturalistic studies of patients with mixed episodes of bipolar disorder. The aim of this study was to examine 10-year outcomes in patients with at least 1 mixed episode. Method: A naturalistic sample of bipolar I disorder patients (n=120), representative of bipolar patients treated in a catchment area of Spain, was followed prospectively for up to 10 years. Outcomes including number (primary study outcome) and severity of episodes, hospitalizations, and suicide attempts were recorded at bimonthly visits. Bivariate and logistic regression models identified factors significantly associated with mixed episodes. The study was conducted from 1994 through 2004. Results: 37% of patients had mixed episodes. Mixed-episode patients had younger mean age at onset compared with the nonmixed group (25.3 vs 30.8 years; P=.025). After adjusting for age at onset, mixed-episode patients had an increased risk of hospitalization compared with the nonmixed group (OR=2.86; 95% CI, 1.09-7.52; P=.033) and more episodes (OR= 1.21; 95% CI, 1.10-1.31; P<.001). Other differences between mixed and nonmixed patients, such as alcohol abuse, psychotic symptoms, and suicidality, were partially mediated by age at onset and were not significantly different after controlling for this variable. Mixed-episode patients with previous suicide attempts had a significantly shorter time to first suicide attempt during follow-up than those without history of suicide attempts (P=.014). Conclusions: Although some factors associated with mixed episodes are mediated by a younger age at onset, the long-term prognosis of mixed-episode patients is worse than patients with nonmixed episodes.