BACKGROUND: For sudden collapse from out-of-hospital cardiac arrest (OHCA) in adults, bystander chest compression-only CPR (COCPR) results in equivalent or better survival compared with conventional CPR with rescue breathing. However, patients who arrest from a non-cardiac cause may also receive COCPR even though rescue breathing might benefit them. We sought to determine whether lay rescuers appropriately provided COCPR or conventional CPR depending on the suspected etiology of OHCA. METHODS: Analysis of a statewide, Utstein-style registry of adult OHCA during a large-scale campaign endorsing COCPR for adults who suddenly collapse. The relationship between bystander CPR (both COCPR and conventional CPR), presumed arrest etiology, and survival to hospital discharge were evaluated. Means and 95% confidence intervals (CI) were calculated. FINDINGS: A total of 7,652 adult OHCAs not witnessed by EMS were analyzed from 10/04 to 09/10. After exclusion of non-lay bystanders and arrests in medical facilities, presumed cardiac etiology and non-cardiac etiology OHCA totaled 4,913 and 1,202, respectively. The presumed specific non-cardiac etiologies were non-cardiopulmonary (e.g. overdose, suicide, subarachnoid hemorrhage) (57%), trauma (27%), and respiratory (16%). All-rhythm survival was 7.0% (CI: 6.3 - 7.8) for cardiac etiologies and 3.3% (CI: 2.3 - 4.3) for non-cardiac (p<0.001). Bystander CPR was provided for 33% (CI: 31.4 - 34.1) of victims with presumed cardiac causes versus 28% (CI: 25.6 - 30.7) for non-cardiac causes (p=0.002). Presumed respiratory causes of OHCA had a much lower proportion of COCPR (8.3%, CI: 4.4 - 12.2) than cardiac causes (18.0%, CI: 16.9 - 19.1, p<0.001). Survival for non-cardiac cases was: 4.4% (CI 0.9 - 7.8) for conventional CPR, 3.5% (CI: 0.9 - 6.0) for COCPR, and 3.1% (CI: 2.0 - 4.3) for no CPR (p = 0.66). CONCLUSIONS: In the setting of a large-scale, statewide public health campaign endorsing lay rescuer COCPR for adults who suddenly collapse and are not breathing normally, bystanders were much less likely to perform COCPR on OHCA victims who might benefit from rescue breathing. Furthermore, when COCPR was performed on presumed non-cardiac cause arrest victims, there was no demonstrable negative impact on survival.