Objective: To identify characteristics associated with provision of bystander CPR in witnessed out‐of‐hospital cardiac arrest cases.Methods: An observational, prospective, cohort study was performed using cardiac arrest cases as identified by emergency medical services (EMS) agencies in Oakland County, MI, from July 1, 1989, to December 31, 1993. All patients who sustained a witnessed arrest prior to arrival of EMS personnel were reviewed.Results: Of the 927 patients meeting entry criteria, the 229 patients receiving bystander CPR were younger: 60.9 ± 14.7 vs 67.9 ± 14.7 years (p < 0.01). Most (76.6%) cardiac arrests occurred in the home. In a multivariate logistic model, only the location of arrest outside the home was a significant predictor of receiving bystander CPR [odds ratio (OR) 3.8; 99% CI 2.5, 5.9]. Arrests outside the home were associated with significantly improved outcome, with 18.2% of out‐of‐home and 8.2% of in‐home victims discharged from the hospital alive (OR 2.5; 99% CI 1.4, 4.4).Conclusion: Patients who have had witnessed cardiac arrests outside the home are nearly 4 times more likely to receive bystander CPR, and are twice as likely to survive. This observation emphasizes the need for CPR training of family members in the authors' locale. This phenomenon may also represent a significant con‐founder in studies of out‐of‐hospital cardiac arrest and resuscitation.