BACKGROUND: The optimal cardiopulmonary resuscitation (CPR) duration for patients with out-of-hospital cardiac arrest (OHCA) remains unclear. We aimed to assess the association between CPR duration and outcome after OHCA. METHODS: This prospective, population-based observational study conducted in Osaka, Japan enrolled 6981 adult patients with non-traumatic witnessed OHCA who achieved return of spontaneous circulation (ROSC) from January 2005 through December 2012. CPR duration was defined as the time of CPR initiation by emergency medical service personnel to the ROSC in pre-hospital settings or after hospital admission. The primary outcome was one-month survival with neurologically favourable outcome (cerebral performance category scale 1 or 2). RESULTS: Overall, median CPR duration was 25min (interquartile range: 15-34) and the proportion of neurologically favourable outcome was 12.5% (875/6,981). The proportion of neurologically favourable outcome among the CPR duration ≥31min group was significantly lower compared with that among the 0-5min group (55.1% [320/581] versus 2.2% [54/2424], adjusted odds ratio [AOR] 0.04; 95% confidence interval [CI] 0.03-0.05 in all patients, 78.4% [240/306] versus 11.4% [30/264], AOR 0.04; 95% CI 0.02-0.06 in the shockable group, 29.1% [80/275] versus 1.1% [24/2160], and AOR 0.03; 95% CI 0.02-0.05 in the non-shockable group). The cumulative proportion for neurologically favourable outcome reached 99% after 44, 41, and 43min of CPR in all patients, the shockable group, and the non-shockable group, respectively. CONCLUSION: The proportion of patients with neurologically favourable outcome declined with increasing CPR duration, but some OHCA patients could benefit from prolonged CPR duration >30min.