Clinical mechanical chest compression studies report diverging outcomes. Confounding effects of variability in hands-off fraction (HOF) and timing of necessary tasks during advanced life support (ALS) may contribute to this divergence. Study site variability in these factors coupled to randomization of cardiopulmonary resuscitation (CPR) method was studied during simulated cardiac arrest prior to a multicentre clinical trial.Ambulance personnel from four sites were tested in randomized, simulated cardiac arrest scenarios with manual CPR or load-distributing band CPR (LDB-CPR) on manikins. Primary emphasis was on HOF and time spent before necessary predefined ALS task (ALS milestones). Results are presented as mean differences (confidence interval).At the site with lowest HOF during manual CPR, HOF deteriorated with LDB-CPR by 0.06 (0.005, 0.118, p=0.04), while it improved at the two sites with highest HOF during manual CPR by 0.07 (0.019, 0.112, p=0.007) and 0.08 (0.004, 0.165, p=0.042). Initial defibrillation was 29 (3, 55, p=0.032)s delayed for LDB-CPR vs. manual CPR. Other ALS milestones trended toward earlier completion with LDB-CPR; only significant for intravenous access, mean difference 70 (24, 115, p=0.003)s.In this manikin study, HOF for manual vs. mechanical chest compressions varied between sites. Study protocol implementation should be simulation tested before launching multicentre trials, to optimize performance and improve reliability and scientific interpretation.