OBJECTIVE To examine the use of routinely collected computerised data in clinical audit. METHODS Retrospective review of all analyses of obstetric practice based on a computerised data system from January 1983 to June 1988. METHODS Maternity department of the regional referral hospital in Oxford. METHODS Congruence with the principles of clinical audit; that is, comparing clinical practice with previously agreed standards and changing practice to meet these standards if necessary. RESULTS Over the five and a half years of the study the data formed the basis of 130 special inquiries into different aspects of obstetric practice. Most inquiries seemed to be aimed only at describing current activities and identifying trends. Genuine clinical audit was rare. Simple audits--for example, concerning induction for pregnancy after term--could be supported by the computerised data, but for detailed and wide ranging audits--for example, reducing antenatal clinic visits for low risk multiparas--the data had to be supplemented from other sources. CONCLUSIONS Routinely collected computerised data enable ongoing clinical audit, but it becomes a reality only when clinicians agree on standards of practice and have a flexible attitude towards change. Even then, genuine clinical audits of obstetric practice demand more detailed and comprehensive data than are generally available on such systems.