The discrepancy between the constancy or increase of the prevalence of depressive symptoms and dysphoria in old age on one hand, and the decrease in the prevalence of the DSM-III diagnoses of major depression and dysthymia on the other, is discussed in light of the most frequent explanatory hypotheses such as memory defects, interpretation of depressive as somatic symptoms, higher risk of institutionalization as well as higher mortality of depressives and a mitigated course of depression in old age. We conclude that higher mortality, mitigation and the rarity of true late-onset depression are arguments for a real decline in prevalence, which occurs in accordance with the decline in all psychiatric disorders that are connected with emotional upheavals and substance ingestion. On the other hand, the connection of depressive states with somatic illness is strengthened, and according to preliminary validation studies, clinically relevant depressive states not reaching the threshold of DSM-III diagnoses may be typical for the depressive psychopathology of old age.