`In the past, I have worked in a research context where if a person was found to meet the criteria for opioid dependence they received treatment, however if even slightly below the criteria of the DSM-IV they would have to look elsewhere for the disorder. This was an ongoing concern for me, as the person who met the criteria was not always the one with the most discomfort, and alternative treatments were not easy for people to find. Largely from this experience, I find that the current categorical approach to classifying people with psychopathology is, at best, an imperfect system, with the main advantage of being convenient when communicating with other professionals. I wonder whether this convenience comes at a serious cost in terms of accuracy, the result of which is an artificial limit to the range of presentations that occur in psychopathology. As the example above illustrates, the particular aspect that I find most problematic is the use of limits for specific symptoms, for example the duration of the symptoms must have been present for it to be classified as a disorder, or even the number of symptoms that must be eliminated. be present. I think it is unlikely that a person who "almost" meets the criteria for a disorder is significantly different from a person with similar symptoms who barely meets the criteria. In private practice these two cases would likely be treated similarly, but in a context where the diagnosis serves as a screening tool, the client who meets the criteria may receive treatment while the other does not. In this case I believe that less specific guidelines, without specific numerical limits, would alleviate many of the problems. A separate, but related, issue lies in the accuracy of the diagnosis in identifying a distinct pattern of pathology. I... half the paper... could be used in therapy. The obvious limitation to this lies in the ease of transmitting information to other professionals. However, I believe that the limitation is not a serious problem, since the usefulness of the information provided by a diagnosis is questionable. However, I recognize a serious advantage to categorical diagnosis in therapeutic education and research. It would be impossible to train students on the full range of presentations that can occur, and having a limited range of diagnoses provides a good basis for conceptual groupings. The ability to use these groupings for research also promotes better training. This is the only area where I believe categorical diagnosis is a requirement. In conclusion, I believe that the ideal system is training using a categorical system, but in the context of real application a more descriptive system.