Six new developments in the classification and diagnosis of functional mental disorders are described. These are: the standardization of data collection, the formulation of operational diagnostic criteria, the multiaxial approach to psychiatric diagnosis, the development of compromise classification systems, the polydiagnostic approach, and, finally, the idea of validating diagnostic concepts instead of taking them for granted. The merits and pitfalls of each of these new developments are discussed. It is concluded that both the traditional categorical diagnostic approach and the introduction of diagnostic compromise systems are an obstacle to progress in psychiatric research. On the other hand, a blind belief in operational diagnostic criteria also has its drawback. Only efforts at standardized data collection, at using a multiaxial and polydiagnostic approach, and regarding psychiatric diagnoses as hypothesis, will contribute to progress in psychiatric research.