The fatigue interlaboratory test campaign organized by the RILEM 182 PEB Technical Committee is presented. Sinusoidal loading tests were performed on the same mixture at a frequency of 10Hz and a temperature of 10 degrees C, for 11 different kinds of tests. They include tension/compression, 2, 3 and 4 point bending, and indirect tension tests. More than 150 fatigue tests were done with data acquisition during the whole duration of the experiment. The analysis with the classical approach, which shows different results following the kind of tests and following the mode of loading, appears as very limited and not intrinsic. Biased or artefact effects (heating, thixotropy) appearing during laboratory fatigue tests seem to have non negligible influence and have to be taken into account to obtain the "true" fatigue property. Other approaches based on damage theory are proposed. Different modellings with damage laws are also presented. As the classical analysis has great limitation and weakness, the damage approach, which is rational, appears as a good option to deal with fatigue of mixes. For the covering abstract see ITRD E117423.