AIM: To describe survival after in-hospital cardiac arrest in relation to the interval between collapse and start of cardiopulmonary resuscitation (CPR). PATIENTS: All patients suffering in-hospital cardiac arrest in Sahlgrenska University Hospital in Göteborg, Sweden between 1994 and 1999 in whom resuscitative efforts were attempted and for whom the interval between collapse and start of CPR was known. METHODS: Prospective recording of various factors at resuscitation including the interval between collapse and start of CPR. Retrospective evaluation via medical records of patients' previous history, clinical situation prior to cardiac arrest and final outcome. RESULTS: Survival to discharge was 33% among the 344 patients in whom CPR was started within the first minute as compared with 14% among the 88 patients in whom CPR started more than 1 min after collapse (P=0.008). The corresponding figures for patients found in ventricular fibrillation was 50 versus 32% (NS); for patients found in pulseless electrical activity 9 versus 3% (NS) and for patients found in asystole 19 versus 0% (NS). Correcting for dissimililarties in the previous history and factors at resuscitation, the adjusted odds ratio and 95% confidence limits for being discharged from hospital when CPR was started within 1 min compared with a later start was 3.06 with 95% confidence limits of 1.59-6.31. CONCLUSION: Among patients with in-hospital cardiac arrest in whom the interval between collapse and start of CPR was known, we found that in 80% of the cases CPR was started within the first minute after collapse. Among these patients, survival to discharge was twice that of patients in whom CPR was started later. These results highlight the importance of immediate CPR after in-hospital cardiac arrest.