For many years scientists and physicians have pondered upon the apparent connection between depressive disorder and diabetes mellitus. Several epidemiologic studies confirm that diabetics have increased incidence of depression, and vice versa. In addition: depressive, non-diabetic patients have several insulin- and glucose-metabolism disturbances, probably exerting a compensatory reaction to the malfunction in the depressed brain as these disturbances are normalised in remission. After the discovery of PET-scanning, such studies have shown that patients with depressive disorder have reduced glucose metabolism in frontal parts of the brain. The present hypothesis regards the PET findings as observations of the primary pathophysiology of depression. Furthermore: two studies of post mortem samples from depressed patients show reduced numbers of astroglia. This is in accordance to the mentioned insulin disturbances, as only astroglia, not neurons, have insulin-sensitive glucose metabolism. Hence: the astroglia, not necessarily the neurons, are proposed to be the type of cells in which the disease resides. Most probably depressive disorder is a multitude of diseases, explaining the apparent multitude of symptoms, and the fact that different patients do respond to different drugs. Therefore: one can only formulate the hypothesis by mentioning a common denominator to these specific malfunctions, namely: disturbed glucose metabolism in the depressed brain. The present paper reviews several findings and proposes that attenuated cerebral glucose metabolism in frontal parts of the brain, in the astroglia, is the cause of depressive disorder.