The definition of a separate nosological entity "panic disorder" in the Diagnostic and Statistical Manual of Mental Disorders, 3rd revision (DSM III) of the American Psychiatric Association triggered a flood of empirical research into this disorder. In this article the literature is surveyed to find out to what extent panic attacks and panic disorder bear specific relationships to measurable variables from the psychosocial and biological spheres. The investigated psychosocial variables are: experiences of separation in childhood; separation anxieties in childhood; sociodemographic correlates as they are known from epidemiological studies; personality factors; life events as possible triggering events of periods with panic attacks; and, finally, socalled "cognitive correlates". The results obtained in this field are just as inhomogeneous as they continue to be in the sphere of biological variables. Of these, the following factors are discussed: genetic factors; neurological and psycho-physiological correlates; the problem of mitral valve prolapse; endocrinal correlates and neurotransmitters. Present findings do not supply conclusive evidence either for or against separation of panic attacks from free-floating anxiety. The relationship between panic attacks and depressive disorders is equally unclear if judged on the basis of the results from empirical research which are far from uniform.