<p><b>Objective.</b> To study associations between cortisol levels, clinical features of depression, its nosological affiliation, and the nature of therapeutic responses in female patients. <b>Materials and methods.</b> Sixty-nine women aged 18–50 years were investigated using clinical-psychopathological, clinical follow-up, psychometric, laboratory, and statistical methods. <b>Results.</b> Seventeen patients were included in the group with depression without signs of resistance and 52 in the group with treatment-resistant depression (TRD). Hypercortisolemia at baseline was observed in 23% and hypocortisolemia in 6%. Failure of cortisol suppression in the low-dose dexamethasone suppression test (LDDST) was detected in 42%. Patients with hypercortisolemia were diagnosed with episodic schizophrenia more often than patients with normal cortisol – three out of 16 (19%) and one out of 49 (2%) respectively (<i>p</i> = 0.043). No differences were found in the clinical and psychopathological structure of depression or the formal assessment of its severity in patients with hypercortisolemia and with normal cortisol levels. Psychometric assessment showed that patients with hypercortisolemia had significantly higher scores for psychomotor retardation (HAMD, individual registration card) than those with normal cortisol (<i>p</i> = 0.009). In the group with negative LDDST, significantly higher scores for agitation (<i>p</i> = 0.045) and adynamia (<i>p</i> = 0.017) were found than in the group with positive LDDST. Patients with impairments to the circadian rhythm of cortisol secretion showed significantly higher anhedonia scores than patients with normal daily cortisol secretion (<i>p</i> = 0.02). Patients with negative LDDST on repeat examination were in a significantly worse condition at follow-up observations than those with positive LDDST (<i>p</i> = 0.04). In patients with good treatment responses, the cortisol level in the LDDST on follow-up assessment was significantly lower than that in nonresponding patients: 34.1 [22.8–65.3] and 80.2 [66.8–120.8] nM respectively (<i>p</i> = 0.02). <b>Conclusions.</b> The frequency of HPA axis dysfunction was higher in patients with depression than in the general population. HPA axis dysfunction appears to be associated with psychomotor retardation, adynamia, agitation, and anhedonia. The absence of cortisol suppression in LDDST is the most reliable laboratory marker of an intractable depressive state. Absolute treatment resistance in depression is associated with HPA axis dysfunction.</p>