This paper presents a study carried out with extensive weigh-in-motion (WIM) data collected on the French main network, involving 3 millions of truck data. The data were collected by three French WIM stations located on highly trafficked highways and motorways, continuously over a whole year. The data were analyzed using a well-known statistical software. The trucks of the traffic flow were classified into almost 20 classes, depending on the silhouette, number and spacing of axles, type of trailer/semi-trailer, axle (steering, driving or regular axle) and wheel and tire (single, twin, wide), and the data were analyzed by class. First, the longitudinal location of the centre of gravity of the payload was calculated and its variability was analyzed in each class. The loading patterns and behavior were analyzed for semi-trailer with tractor, rigid truck and rigid truck with trailer. In a second step, it was shown that the axle load distribution was mostly multimodal, and the modes have been determined by axle rank and truck class. A maximum of likelihood method was used to do this estimation. The conclusions of these investigations are some recommendations on a better truck loading, on the risk of overloads, on a more accurate truck aggressiveness assessment, and may lead to optimization policies.