Introduction: Bipolar disorder (BD) often starts in childhood or adolescence. There is considerable scepticism outside the United States over the validity, stability and prevalence of BD in children and adolescents. Persistence of course lends support to the validity of a diagnosis. Objectives: To describe the longitudinal course of pediatric BD in a Spanish sample over a median follow-up period of 2.6 years and to examine risk factors associated with outcome. Methods: We retrospectively reviewed the medical records of all children and adolescents (N = 38) with DSM-IV-TR BD-I, II and NOS evaluated in the Child and Adolescent Psychiatry Unit, University of Navarra (Pamplona, Spain) from 1999 to 2005. We used the NIMH Lifetime Mood Chart and the Clinical Global Impression-Severity Scale to assess clinical course. Results: 79% (N = 30) were boys and 21% (N = 8) were girls; 44.7% (N = 17) had BD-I, 5.3% (N = 2) BD-II, and 50% (N = 19) BD-NOS. Median (inter-quartile range: IQR: Q25; Q75) age at diagnosis was 13.9 (10.64; 15.84). Median follow-up period was 2.6 years (0.91; 3.66). Mean percentage of time in an episode was 46.17% (23.36; 75.26), and it was longer in younger children (p < 0.05). 2.6% had rapid cycling. At the end of follow-up, only 47% achieved remission or recovery. Younger children showed a worse treatment response (p < 0.05). We found higher rates of hospitalization in children with ADHD (21%) (p < 0.05). Conclusion: Children with BD had a chronic course with little interepisodic recovery. BD can be diagnosed in children using DSM-IV-TR criteria. An early age of onset and ADHD comorbidity are risk factors for worse prognosis.