The objective of the trials was to estimate in different products (matrices) the components of the measurement uncertainty (MU) linked to taking the test portion (in the test sample) and to the preparation of the initial suspension. The components linked to the sample storage and transportation are not taken into account since the trials are performed on different samples in each laboratory. These trials were designed to calculate, in addition to the MU component strictly linked to the laboratory (from the initial suspension to the counting/confirmation of colonies and expression of results), the MU components linked to the initial steps of the analysis, from the test portion to the preparation of the initial suspension. It was supposed that this part of MU would not be (mainly) depending on the laboratory which estimates it, but rather on the type of matrix considered1. Thus we thought each laboratory would have been able to estimate its global MU, which would have comprised a common part, the MU on the test portion and the preparation of initial suspension (typical value defined per type of matrix), and an individual part, the MU based on the intra-laboratory standard-deviation.