The dynamic stiffness of a resilient material used under a floating floor is often used to predict the improvement of the impact sound pressure level, L. It is also used to compare products. The measurement accuracy of this parameter is therefore essential. Unfortunately, the comparison between the predicted and the measured L results shows quite high deviations which could be attributed, in part, to an incorrect estimation of the dynamic stiffness. It is now accepted by all European laboratories that the measurement procedure described in the standard ISO 9052-1 should be reviewed. This paper proposes a first step in the improvement of the measurement setup by taking into account the actual contribution of the dynamic stiffness of the air enclosed in the materials on the total dynamic stiffness. A new setup is proposed and some results are presented for products with different airflow resistivities.