We examined successively the different possible approaches to pathological anxiety. In clinical practice, it is easy to distinguish, in qualitative and quantitative terms, between neurotic and psychotic anxiety, the difference being obvious to any practising psychiatrist. A psychopathological approach, however, raises some doubt as to whether the answer is so obvious, and leads us to pose two questions: 1) Is it still possible to have a unitary conception of pathological anxiety, commonly considered as a central element to all forms of psychopathological expression? 2) Since anxiety is an emotion which can be felt by any normal as well as pathological individual, should we consider anxiety to be different when it is expressed in different states - neurosis, psychosis or even border states - by different types of symbolization , depending on the organization of the state? The present advances in pharmacoclinical research, using pharmacological agents as modifying agents for behaviour patterns with specific action mechanisms, enable us to carry the debate further by distinguishing clinical phenomena and neurobiological effector mechanisms. It also shows that within the various established clinical types, several pathological syndromes may be involved, each having different symptoms and its own distinct evolution, and requiring different, more specific treatment. This leads us to rethink the classical distinctions, to refine our semiology, and to familiarize ourselves with the use of chemotherapy, particularly as a "behaviour analyzer", able to teach us something about the different types of anxious behaviour and about the neurobiological mechanisms which are specifically modified by these pharmacological agents.