The paper reviews available epidemiological evidence for the existence of and the implications of comorbidity of anxiety and depressive disorders and mixed anxiety-depressive (MAD) disorders. Using epidemiolological evidence of prevalence and incidence and data relating to time-course of illness, risk factor and outcome, it is concluded: (1) that anxiety-depression comorbidity is quite frequent in epidemiological and clinical settings throughout the world; (2) this comorbidity is diagnosis-specific and is associated with increased vulnerabilities and risks as well as poorer outcome and marked disabilities; and (3) no such evidence was found for MAD disorders. Contrary to what was predicted, the prevalence of MAD disorders was quite low even when using the more recent criteria of the Diagnostic and Statistical Manual of Mental Disorders, 4th edition. (4) Furthermore, there was quite a heterogeneous pattern in terms of risk, severity and outcome making it questionable whether this disorder, as currently defined, is a clinical entity. These findings are discussed in terms of two perspectives, the 'lumpers' with their dimensional view and the 'splitters' with their categorical view. It is concluded that although comorbidity of threshold anxiety and depressive disorders seems to be an important phenomenon, no such evidence is provided for MAD disorders. Copyright 2001 John Wiley & Sons, Ltd.