To gain more insight into decision making around the termination of resuscitation (CPR), we studied factors which influenced the time before discontinuing resuscitation, and the criteria on which those decisions were based. These criteria were compared with those of the European Resuscitation Council (ERC) and the American Heart Association (AHA). For this study, we reviewed the audiotapes of resuscitation attempts in a hospital. A total of 36 attempts were studied, involving 27 men and nine women, mean (S.D.) age 64 (18) years. A total of 19 patients received resuscitation on general wards, and 17 in the emergency room after an out-of-hospital circulatory arrest. The median interval time (range) from start to termination was 33 min (8-81 min). Results from multiple linear regression showed that a delay greater than 5 min in first advanced life support measures, drawing a sample for biochemical analysis, and the patient's response shown by return of spontaneous circulation were independently associated with the time of terminating resuscitation. The team used a number of criteria which can be found in the guidelines of the ERC and the AHA, but also used additional criteria. The ERC and the AHA criteria were not sufficient to cover all termination decisions. We conclude that the point in time to terminate resuscitation is not always rationally chosen. Updating of the current guidelines for terminating resuscitation and training resuscitation teams to use these guidelines is recommended.