Summary Introduction Previous studies have documented poor quality CPR during real life resuscitation attempts. This study investigated whether poor quality CPR during advanced life support training could be contributing to poor performance in clinical practice. Methods Observational study of quality of CPR during advanced life support training courses before and after the implementation of the European Resuscitation Council Guidelines 2005 into the ALS course. The quality of chest compressions were downloaded from a manikin and direct observations of no-flow time; pre-shock pauses were recorded. Results 94 cardiac arrest simulations were studied (46 before implementation of Guidelines 2005 and 48 after). Delays in starting CPR, inadequate compression depth, prolonged interruptions of chest compressions and excessive pre-shock pauses were identified. The introduction of Guidelines 2005 resulted in improvements in the number of compressions given per minute and a reduction in no-flow time and duration of pre-shock pauses, but overall the quality of CPR performed during the ALS course remained poor. There was little evidence of performance improving over successive simulations as the course progressed. Conclusion The implementation of Guidelines 2005 into the ALS course appear to have improved the process of CPR by reducing no-flow time during simulated CPR. However, the quality of CPR during ALS training remains sub-optimal. Delays in starting CPR, inadequate compression depth, excessive interruptions in chest compressions and prolonged pre-shock pauses mirror observations from clinical practice. Strategies to improve CPR performance during ALS training should be explored and potentially may result in improvements in clinical practice.