Registers on people with learning disability are widely held (Cubbon & Malin, 1985). They have been set up in a variety of ways, hold differing types of data, and are put to various uses. Service planning and management is a common function for which register data may be required. However, a literature search revealed that no accuracy studies had been undertaken on such registers, apart from a reliability study on the Wessex Mental Handicap Register developed by Kushlick et al (May et al, 1982). In Wandsworth, an inner city health authority, the Register for People with Learning Difficulties is jointly funded and managed by the health authority and local social service department. The register holds data on just under 1,000 people who use services for people with learning disabilities, and the data are updated annually by questionnaires sent out by the register organiser. The questionnaire includes enquiry into demographic information, functional ability and service use. It also notes the presence or absence of additional complications, including psychiatric disorder and behaviours presenting specific management problems. This article describes a pilot study of the accuracy of the psychiatric and behaviour problem data held on the Wandsworth Register.