In the acute care model (people get sick, see doctor, get remedy, get well) it is always presumed that people 'see' a clinician, even if the clinician is only connected by a video link. Medicare is based on this model. This piece is about treating people you never see, specifically about the web-based treatment of common internalising menta] disorders, e.g. major depression, generalised anxiety disorder, panic disorder and social phobia.The traditional model is to see people, listen to the history of symptoms and then examine the person for the signs associated with illness. Telephone counselling for anxiety and depression by trained professionals ' or by trained lay people2 has been shown to produce benefits commensurate with face-to-face treatment, so one doesn't always have to see the person.The internet is the greatest collection of freely available medical information that has ever been constructed. There is no doubt that it can inforni - but sometimes it misinforms and some clinicians are wary of the patient with an internet print out. Surely sensible information does help people with chronic disorders understand and manage themselves. Patient education on a face-to-face basis has been shown to improve outcome and reduce inappropriate health service utilisation.3'4 We are unaware of internet education having been shown to do this, but it seems likely that it would be effective.The larger question is whether the internet can deliver treatment, i.e. an intervention in the course of an illness that is (a) superior to the value of good information on disease management and (b) equivalent in benefit to good face-to-face treatment. Clearly the supply of prescription medication to someone you have never seen is not practical, as is surgery and physical manipulations, but psychotherapy does seem to be a possibility. Marks et air reviewed a large number of systems to provide computer aided psychotherapy largely, but not always, supervised by a clinician. Moodgym6 was an early example of a web-based intervention that was effective in reducing depression in the population with a standardised mean difference (or effect size) to the control group of 0.35; although it was eventually found not to be superior to good information about effective treatments for depression. Nevertheless, the health gain from open use of this free program is likely to have been considerable.Four internalising disorders (major depression, generalised anxiety disorder, panic disorder and social phobia) together contribute 6% (150000 disability-adjusted life years lost (DALYs) in Australia) to the burden of human disease.7'8 Although these disorders present differently they share similar vulnerability or risk factors and are probably aspects of the same underlying disorder. The nature of these disorders is well established. Twin data show that the genetic risk associated with major depression is substantially shared with generalised anxiety disorder,10 and to a lesser extent with the phobias. Neuroticism, a personality trait sometimes termed negative affectivity, also explains some of the common genetic risk of the internalising disorders.'12 High scores for neuroticism precede the development of these disorders.'3'4 What does this genetic liability produce? Mineka el a/.15 conclude that anxiety and depressive disorders are associated with automatic attentional biases for emotion-relevant material, and with negative judgmental or interpretive biases. Indeed the anxiety and depress i ve di sorders appear to be risk factors for each other.16Cognitive behaviour therapy (CBT) and selective serotonin reuptake inhibitors (SSRIs) separately or together can be effective treatments in all the internalising disorders. However, treatment with SSRIs can be disappointing; effect sizes in research studies are only moderate and improvement stops when the medication is stopped. In practice, patient adherence is poor. Treatment with CBT is better; effect sizes are marginally larger than with drugs and the improvement lasts after treatment ends. …