Background “Hostile depression” has unofficially long been described as a depressive subtype, but since DSM-III, the affect has been made a defining characteristic of borderline personality disorder. The related affect of irritability in DSM-IV-TR subsumes various hostile nuances and is included in the stem question for mood disorders—especially for hypomanic episodes; in children, it is nonetheless a sign of depression. Then, there is the unofficial more general concept of depression with anger attacks, until recently ostensibly a “unipolar” (UP) disorder. A veritable tower of Babel indeed. In the present analyses, our aim was to extend previous research on irritable-hostile depression to more specific parameters of bipolarity and depressive mixed state (DMX). Methods Consecutive 348 bipolar-II (BP-II) and 254 unipolar (UP) major depressive disorder (MDD) outpatients (off psychoactive agents, including substances of abuse), were interviewed with the Structured Clinical Interview for DSM-IV, the Hypomania Interview Guide, and the Family History Screen. Borderline personality, a confounding variable, rare in the FB setting, was excluded. Irritability was defined according to DSM-IV-TR, which includes various features of hostility and anger. Depressive mixed state (DMX) was defined as a major depressive episode (MDE) plus three or more concurrent intradepressive hypomanic symptoms, whether it occurred in BP-II or MDD. Results MDE with irritability was present in 59.7% (208/348) of BP-II and in 37.4% (95/254) of MDD ( p=0.0000). In BP-II, MDE with, versus MDE without, irritability had significantly younger index age, higher rates of axis I comorbidity, atypical depressive features, and DMX. Upon logistic regression, we found a significant independent association between BP-II MDE with irritability and DMX. In UP, MDE with, versus without, irritability had significantly younger age and age at onset, higher rates of atypical depression, DMX, and bipolar family history. Logistic regression revealed a significant independent association between MDE with irritability and DMX. Given that we had excluded patients with borderline personality, the high prevalence of irritable-hostile depressives in this outpatient population means that hostility cannot be considered the signature of that personality. Factor analysis revealed independent “psychomotor activation” and “irritability-mental activation” factors. Odds ratios of irritability for DMX were highest in the “UP” MDD group (=12.2); for predicting DMX, irritability had the best psychometric profile of sensitivity of 66.3% and a specificity of 86.1% for this group as well. Limitation We did not use specific instruments to measure irritable, hostile, and angry affects. Conclusions These analyses show that irritable-hostile depression is distinct from agitated depression. Whether arising from a BP-II or MDD baseline, irritable-hostile depression emerges as a valid entity with strong links to external bipolar validators, such as bipolar family history. Irritable-hostile phenomenology in depression appears to be a strong clinical marker for a DMX. Irritable-hostile depression as a variant of DMX deserves the benefit of what seems to work best in practice, i.e., anticonvulsant mood stabilizers and/or atypical antipsychotics. Formal treatment studies are very much needed.