During active compression-decompression cardiopulmonary resuscitation (ACD-CPR), the rescuer applies traction to the chest between compressions. Under experimental conditions, cardiac output increases, possibly through accentuated intrathoracal pressure fluctuations. ACD-CPR requires specific training and may be more complex to perform than standard CPR. The aim of this study was to characterize ACD-CPR performance compared with standard CPR by emergency care providers; in terms of decompression force, compression rate, depth and duration. Thirty-three ambulance paramedics were studied while performing standard CPR and ACD-CPR with the Ambu Cardiopump on a specially designed transducer-equipped manikin 9 months following initial training. The order of CPR performance was determined randomly by cross-over design. Performance data were recorded by a computer. The 2-min average active decompression force was 9.3 kg (interquartile range 2.5-15.3 kg) and six subjects (18%) met the manufacturers recommendation of 10-15 kg. External chest compression (ECC) rate decreased from 85 (70-101) to 76 (63-88) min-1 (P < 0.001), ECC depth decreased from 54 (50-58) to 45 (39-48) mm (P < 0.001) and compression duration from 40 (35-45) to 31% (28-33%) (P < 0.001) upon change from standard CPR to ACD-CPR. We conclude that the recommended level of decompression force was achieved by less than one fifth of study subjects. ACD-CPR when compared with standard CPR causes a consistent and significant reduction of compression rate, depth and duration. These are all factors of possible clinical significance. Training in ACD-CPR should address this issue, with special emphasis on optimal decompression force and ECC rate.