BACKGROUND: Despite the well-known importance of witnessed arrest and bystander cardiopulmonary resuscitation (CPR) for out-of-hospital cardiac arrest outcomes, previous studies have shown significant statistical inconsistencies. We hypothesized an interaction effect and conducted stratified analyses to investigate whether witnessed arrest is more important than bystander CPR. METHODS: This study enrolled patients with out-of-hospital cardiac arrest between January 2010 and December 2022 in 3 emergency medical service (systems in Taiwan). Data were extracted from emergency medical service dispatch reports, including patient characteristics, witnessed arrest, bystander CPR, time for each dispatch, and prehospital interventions. The outcome measure was prehospital return of spontaneous circulation (ROSC). Patients were categorized into 4 groups: witnessed and bystander CPR present (W+B+), witnessed present but bystander CPR absent (W+B-), witnessed absent but bystander CPR present (W-B+), and witnessed and bystander CPR absent (W-B-). Multiple logistic regression on prehospital ROSC were performed in the 4 subgroups separately. RESULTS: A total of 14 737 patients with out-of-hospital cardiac arrest were identified, of whom 977 (6.6%) achieved prehospital ROSC. The W+B+ group exhibited the highest prehospital ROSC rate (14.0%). Stratification confirmed a statistically significant interaction between witnessed arrest and bystander CPR. Defibrillation, endotracheal intubation, and epinephrine administration were significantly associated with prehospital ROSC in all subgroups. Most explanatory variables significant in the witnessed arrest group were adjusted for in the nonwitnessed arrest group. Younger age was associated with prehospital ROSC only in the W+B+ group. CONCLUSIONS: Witnessed arrest and bystander CPR may interact to predict prehospital ROSC in out-of-hospital cardiac arrest, with witnessed arrest likely having more significant impact on outcomes.