OBJECTIVE--To develop and test a method for routine data collection to observe current practice in outpatient pain clinics. DESIGN--Prospective questionnaire survey completed by consultants on each patient seen during October 1989 to May 1990. SETTING--Outpatient pain clinics of five teaching and five district general hospitals in Scotland and northern England. MAIN OUTCOME MEASURES--Number of new referrals and their source, and characteristics of pain at presentation. RESULTS--4354 forms were completed by 21 consultants over 29 weeks, corresponding to 2241 patients, of whom 981 were new referrals. The proportion of consultations at which new referrals were seen varied among the 10 clinics from 15% to 34%. The difference could not be accounted for by type of hospital. Sources of new referrals varied widely between the clinics (for example, range 22% to 78% for general practitioner referrals), as did new referrals by the type of pain (range 10.8% to 55.2% for low back pain, 10.5% to 32.5% for pain associated with surgery). However, these differences in types of patients seen could not be accounted for by variations in referral patterns among clinics. Problems identified in performing an audit of outpatients included the difficulty of obtaining firm diagnoses, the need for a method to link successive patient contacts, and the complexity of the presenting problem in many patients. CONCLUSIONS--It is possible to collect data for audit routinely in outpatient clinics. Observation of current practice in the clinics suggested possible unmet need or inappropriate management, which may require changes in practice.