To determine the level of knowledge achieved by patients who have read a simple information sheet on gastroscopy and flexible sigmoidoscopy, and to determine the levels of information required by patients and solicitors specializing in clinical negligence. Self-administered questionnaires were sent to 516 patients in Leicester and 79 solicitors specializing in clinical negligence in England and Wales. The main objective outcome measures were the correct scores measured in a simple test of the content of information sheets about gastroscopy and flexible sigmoidoscopy. Other scores were based on the opinions of patients and solicitors on the type of information patients should receive and the levels of risk about which they should be informed. Of the clinical negligence specialists, 75%, compared with 44% of patients, felt that consent to procedures should be obtained 2 weeks before the test. In addition, 86% of solicitors felt that patients needed to be told about tests on at least two occasions and favoured booklets and videos. Both 48% of solicitors and 38% of patients felt that patients should be told of very uncommon risks, with 16% of clinical negligence specialists expecting patients to be told of risks of 1 in 1,000,000. The majority of patients (98%) and solicitors (95%) felt that patients should be formally tested as to whether they understand what they are told in the consent process. In an initial study of an information leaflet, South Asian patients had significantly poorer scores than English patients on formal testing of their knowledge of the leaflet's content. The poorest response (61% correct) was to a question including the word 'indigestion', which had been omitted from the information leaflet. Its inclusion led to a significant increase in the number answering the question correctly (96%). The positioning of a word in the list of reasons for doing the test was also important. In the flexible sigmoidoscopy leaflet, the word 'anaemia' was placed in the middle of a list of reasons for doing the test; only 85% answered the question correctly. In the sample of patients who received the modified gastroscopy leaflet, recall at 6 months was significantly impaired with most loss of memory being related to the purpose of the test. Consent needs to be supported by easy-to-read information and the patients' understanding needs to be formally tested. Important concepts must be included in this information as well as any uncommon risks of the procedure. In order to ensure that the information can be understood, the text should be reviewed by an experienced educationalist. If the text is to be used in a multicultural setting, it is important to ensure that patients for whom English is not their first language can easily understand it. The movement away from 'informed consent' towards an 'informed decision' process should facilitate these improvements.