All prehospital resuscitations performed by emergency physicians in the city of Bochum, Germany, were recorded and evaluated prospectively from 1 August 1989 to 30 September 1990. Initially successful cardiopulmonary resuscitation (CPR) was achieved in 33.8% (alive at admission), and definitive success in 10.4% (discharged alive). Of the patients who presented with cardiac arrest before the arrival of the emergency physician, 28.4% could be resuscitated initially and 7.6% survived definitively. In patients who suffered circulatory arrest on or after the arrival of the emergency physician, the initial success rate of CPR was 51.7%, and the definitive success rate 11.1%. Although the initial success rate was significantly more favourable, the definitive outcome did not differ statistically between these two.groups. Two time periods were compared. During the first period of 5 months only one base with two ambulances staffed with emergency physicians was available. During the second period of 9 months an additional base with a physician-staffed ambulance was established. This reorganization resulted in the reduced call to arrival time falling from 8.5 ± 2.4 min to 7.6 ± 2.4 min (univariate variance analysis:f=8.89, d.f.=1.31,p<0.01). This decrease, however did not improve either the initial or the definitive success of CPR. From these results we conclude that further improvement of prehospital resuscitation can only be achieved to a small extent by reducing the call to arrival time of ambulances staffed with emergency physicians. Improvement is more likely to be seen when immediate resuscitation is performed by bystanders present at the scene.