Background. This paper presents the first results of a two-stage psychiatric population survey, which uses a new method of directly evaluating needs for specific psychiatric treatment and the extent to which they have been met.

Method. The sample was drawn at random from the population of an area of inner south London with high levels of deprivation. Seven hundred and sixty subjects aged 18–65 completed the GHQ-28. All those scoring >5 and half of the rest were invited to take part in the second stage, comprising measures of mental state (SCAN), social role performance (SRPS), life events and difficulties (LEDS) and a Treatment Inventory. This information was used to rate the community version of the Needs for Care Assessment (NFCAS-C).

Results. In all, 408 subjects were interviewed in the second stage. The weighted 1 month prevalence of hierarchically ordered ICD-10 psychiatric disorders was 9·8%, the 1 year prevalence 12·3%. The equivalent prevalences for depressive episode were 3·1% and 5·3% respectively, while those for anxiety states were both 2·8%. At interview nearly 10% of the population were identified as having a need for the treatment of a psychiatric condition. This rose to 10·4% if the whole of the preceding year was assessed. Less than half of all potentially meetable needs were met. There was only partial overlap between diagnosis and an adjudged need for treatment.

Conclusion. A majority of people with mental health problems do not have proper treatment; given more resources and greater public and medical awareness, most could be treated by family doctors.