Despite of the changes of out-of-hospital cardiac arrest (OHCA) survival rise when bystander CPR is provided, this was only conducted in about 23% of OHCA patients in Korea in 2018. Police officers acting as first responders have a high chance of witnessing situations requiring CPR. We investigated long-term effects on CPR quality between chest compression-only CPR training and conventional CPR training in police officers to find an efficient CPR training method in a prospective, randomized, controlled trial. Police officers underwent randomization and received different CPR training. With the Brayden Pro application, we compared the accuracy of CPR skills immediately after training and the one after 3 months. Right after training, the conventional CPR group presented the accuracy of the CPR skills (compression rate: 74.6%, compression depth: 66.0%, recoil: 78.0%, compression position: 96.1%) and chest compression-only CPR group presented the accuracy of the CPR skills (compression rate: 74.5%, compression depth: 71.6%, recoil: 79.2%, compression position: 99.0%). Overall, both groups showed the good quality of CPR skills and had no meaningful difference right after the training. However, three months after training, overall accuracy of CPR skills decreased, a significant difference between two groups was observed for compression position (conventional CPR: 80.0%, chest compression only CPR: 95.0%). In multiple linear regression analysis, three months after CPR training, chest compression-only CPR training made CPR skills accuracy 28.5% higher. In conclusion, police officers showed good-quality CPR right after CPR training in both groups. But three months later, chest compression-only CPR training group had better retention of CPR skills. Therefore, chest compression-only CPR training is better to be a standard training method for police officers as first responders.