There has been a growing interest in differentiating bipolar depression from unipolar depression, in terms of both phenomenological characteristics and biomarkers. This is more than a question of esoteric academic interest, as any such differentiating features would be of major clinical pertinence. They could assist in the not uncommon clinical scenarios of: first, assisting in determining the likely diagnosis in a depressed patient with uncertain or ambiguous evidence for prior episodes of hypomania or mania and, second, predicting whether a young person with a major depressive episode will go on to develop bipolar disorder or unipolar depression. Further, at a more theoretical level, if there are distinct phenomenological differences between depression in bipolar disorder and unipolar depression, that is likely to indicate a distinct underlying biology between these two disorders. Here, we review studies addressing such putative distinctions, with a focus on the probabilistic approach to the diagnosis of bipolar depression. We argue that future editions of classificatory systems such as diagnostic and statistical manual (DSM) and international classification of diseases (ICD) need to reflect the growing strong evidence for dimensional distinctions between bipolar and unipolar depression.