Background The outcome of out-of-hospital cardiac arrest (OHCA) with a non-shockable rhythm is poor. For patients found in asystole or pulseless electrical activity (PEA), recent guidelines or rules that may be used include “do not attempt to resuscitate” (DNAR) guidelines from Helsinki, discontinuing resuscitation in the guidelines of the European Resuscitation Council and a clinical prediction rule from Canada. We compared these guidelines and the rule using a large Scandinavian dataset. Materials and methods The Swedish Cardiac Arrest Registry includes prospectively collected data on 44 121 OHCA patients. We identified patients with asystole or PEA as the initial rhythm and excluded cases caused by trauma or drowning. The specificities and positive predictive values (PPVs) were calculated for the guidelines, and the clinical prediction rule for comparison. Results A total of 20 484 patients with non-shockable rhythms were identified; 85% had asystole and 15% PEA. The overall survival to hospital admission was 9% (n = 1.861) and 1% (n = 231) were alive at 1 month from the arrest. The specificity of the Helsinki guidelines in identifying non-survivors was 71% (95% confidence interval (CI): 65–77%) and the PPV was 99.4% (95% CI: 99.3–99.5), while the corresponding values for the European Resuscitation Council (ERC) was 95% (95% CI: 91.3–97.5) and 99.9% (95% CI: 99.9–99.9) and, for the prediction rule, 99.1% (95% CI: 96.7–99.9) and 99.9% (95% CI: 99.9–100.00), respectively. Conclusion In this comparison study, the Helsinki DNAR guidelines did not perform well enough in a general OHCA material to be widely adopted. The main reason for this was the unpredicted survival of patients with unwitnessed asystole. The clinical prediction rule and the recommendations of the ERC Guidelines worked well.