BACKGROUND--In September 1990 a working party of the British Pacing and Electrophysiology Group recommended the routine use of physiological pacing systems in patients with bradycardia. An audit of the impact of these recommendations on pacemaker prescription in the Freeman Hospital between March 1990 and August 1991 has previously been reported. This paper considers the effect of that internal audit on subsequent pacemaker prescription from September 1991 to February 1993. PATIENTS AND METHODS--The records of 1228 patients who underwent first pacemaker implantation at the Freeman Hospital between March 1990 and February 1993 were audited retrospectively. The patient's age, indication for pacing, pacing mode, and cost of the generator and leads were recorded. The indications for pacing were sinus node dysfunction (24.9%) (including patients with coexistent sinus node dysfunction and atrioventricular block), atrioventricular block (57.2%), atrioventricular block and atrial fibrillation (12.2%), and carotid sinus syndrome (5.7%). There was an increase in carotid sinus syndrome (2.7% to 8.1%) between the two study periods but no other differences in the distribution of case mix or characteristics of patients. The cost of the working party's recommended optimal pacing mode was calculated from multiplication of the mean cost of the recommended unit over the second half of the study period by the number of patients who would have received that unit. RESULTS--Between March 1990 and August 1991 atrial pacing for sinus node dysfunction (AAI, AAIR, DDD, or DDDR) increased by 138% (from 25.0% to 59.6%), mainly because of increased use of AAI mode. Physiological pacing for atrioventricular block (DDD or VVIR) increased by 41% (from 17.0% to 24.0%), and VVIR pacing for atrioventricular block with atrial fibrillation increased by 111% (from 10.5% to 22.2%). After the internal audit (that is, between September 1991 and February 1993), physiological pacing for atrioventricular block increased by a further 126% (from 24.0% to 54.2%). Sixty three per cent of this increase was in the first six months after the internal audit. Pacemaker prescriptions in sinus node dysfunction and atrioventricular block with atrial fibrillation were unchanged (59.6% physiological pacing for sinus node dysfunction and 22.2% v 27.3% VVIR pacing for atrioventricular block with atrial fibrillation). These changes in practice were accompanied by an increase in the age of patients receiving physiological units. Costs of pacemaker hardware for the final six months of the audit (excluding carotid sinus syndrome) increased by 38% over the costs that would have accrued had pacing policy remained the same as for the initial six month period before the circulation of the recommendations of the working party. Adoption of the guidelines of the working party in full would lead to a further 66% increase in the costs of hardware. CONCLUSIONS--The principal effect of the recommendations of the working party alone was increased use of AAI pacing for sinus node dysfunction, with little change in the costs of hardware. The internal audit was followed by an increase in physiological pacing for atrioventricular block, and this has had important financial consequences. Internal audit was followed by closer adherence to the recommendations of the working party.