It is possible to make certain a priori speculations as to prognostic criteria based on classification and what that implies. We know that pure paranoia is not a deteriorating psychosis--that it does not necessarily preclude the possibility of considerable social usefulness--and that it grades off almost imperceptibly into dementia praecox. The features differentiating these two diseases should therefore supply us with data for determining the prognosis. A case undoubtedly, praecox, which shows markedly the differential features of paranoia, should have a proportionately better outlook. In a vague way our common sense uses this standard when it makes us "feel" that the case will have a long course which shows a relatively well retained personality in conjunction with praecox symptoms. But "feelings" are hardly objective criteria. What symptoms may we make use of? We may say that the praecox patient as opposed to the paranoia has a poverty or inappropriateness of affect, a scattering of thought and a lack of systematization in his delusions. The weakness of will on which Kraepelin lays so much stress may be included, though that can probably be derived from the scattering of thought. What of these symptoms may be analyzed for our purpose? Affect changes and dissociation in the stream of thought are themselves signs of the deterioration we wish to predict; to make use of them we should have at hand some theory as to the relation between their quality and quantity, and that we have not. There remains the content of the psychosis, a definitely objective material with which to work. This is naturally a big problem--almost as wide as insanity itself--and one brief communication cannot pretend to solve it. What we wish to do is merely to put forward tentatively the claim of one type of delusion formation to prognostic value.