The quality of cardiopulmonary resuscitation (CPR) plays a crucial role in saving lives from out-of-hospital cardiac arrest (OHCA). Previous studies have identified sub-optimal CPR quality in the prehospital settings, but the causes leading to such deficiencies were not fully elucidated. This prospective study was conducted to identify operator- and ambulance-related factors affecting CPR quality during ambulance transport; and to assess the effectiveness of mechanical CPR device in such environment. A digital video-recording system was set up in two ambulances in Taipei City to study CPR practice for adult, non-traumatic OHCAs from January 2005 to March 2006. Enrolled patients received either manual CPR or CPR by a mechanical device (Thumper). Quality of CPR in terms of (1) adequacy of chest compressions, (2) instantaneous compression rates, and (3) unnecessary no-chest compression interval, was assessed by time-motion analysis of the videos. A total of 20 ambulance resuscitations were included. Compared to the manual group (n=12), the Thumper group (n=8) had similar no-chest compression interval (33.40% versus 31.63%, P=0.16); significantly lower average chest compression rate (113.3+/-47.1 min(-1) versus 52.3+/-14.2 min(-1), P<0.05), average chest compression rate excluding no-chest compression interval (164.2+/-43.3 min(-1) versus 77.2+/-6.9 min(-1), P<0.05), average ventilation rate (16.1+/-4.9 min(-1) versus 11.7+/-3.5 min(-1), P<0.05); and longer no-chest compression interval before getting off the ambulance (5.7+/-9.9s versus 18.7+/-9.1s, P<0.05). The majority of the no-chest compression interval was considered operator-related; only 15.3% was caused by ambulance related factors. Many unnecessary no-chest compression intervals were identified during ambulance CPR, and most of this was operator, rather than ambulance related. Though a mechanical device could minimise the no-chest compression intervals after activation, it took considerable time to deploy in a system with short transport time. Human factors remained the most important cause of poor CPR quality. Ways to improve the CPR quality in the ambulance warrant further study.