The quality of mechanical CPR (M-CPR) was compared with manual standard CPR (S-CPR) and active compression-decompression CPR (ACD-CPR) performed by paramedics on the site of a cardiac arrest and during manual and ambulance transport. Each technique was performed 12 times on manikins using teams from a group of 12 paramedic students with good clinical CPR experience using a random cross-over design. Except for some lost ventilations the CPR effort using the mechanical device adhered to the European Resuscitation Council guidelines, with an added time requirement of median 40 s for attaching the device compared with manual standard CPR. Throughout the study, in comparison with mechanical CPR the quality of CPR with either manual method was significantly worse. In particular, there were considerable individual variations during stretcher transport. With S-CPR and ACD-CPR the median compression times were 38 and 31%, significantly lower than the recommended 50%, and 46-98% of the decompression efforts with ACD-CPR were too weak, particularly during transport on the stairs. With both manual methods, there were no significant differences in the CPR effort between the site of the arrest and the ambulance transport. However, compression rates were reduced and became more erratic during stretcher transport to the ambulance. When walking horizontally, a median of 19% of S-CPR compressions and 84% of ACD-CPR compressions were to weak. On the stairs, 68% of S-CPR compressions and 100% of ACD-CPR compressions were too weak. In conclusion, when evaluated on a manikin, in comparison with manual standard and ACD-CPR, mechanical CPR adhered more closely to ERC guidelines. This was particularly true when performing CPR during transport on a stretcher.