This article outlines a provisional evidence-based approach to the assessment of pediatric bipolar disorder (PBD). Public attention to PBD and the rate of diagnosis have both increased substantially in the past decade. Accurate diagnosis is crucial to avoid harm due to mislabeling or unnecessary medication exposure. Because there are no proven efficacious or effective treatments for PBD, the role of assessment is heightened to demonstrate efficacy in individual cases as well as to identify cases for participation in clinical trials. This review discusses (a) the state of psychopathology research regarding PBD; (b) the likely base rate of PBD in multiple clinical settings; (c) the diagnostic value of family history; (d) challenges to differential diagnosis, including comorbidity and symptom overlap with other diagnoses, shortcomings in contemporary assessment methods, and the cyclical nature of PBD; (e) practical methods for improving diagnosis, focusing on the most discriminative symptoms, extending the temporal window of assessment to capture mood changes, and using screening tools within an actuarial framework; and (f) monitoring response to treatment using a variety of assessment methods. Twelve recommendations are offered to move toward an evidence-based assessment model for PBD.