The knowledge of loads applied to bridges has to be enhanced in order to actualize national and international codes, like Eurocodes. The nature of traffic loads is extremely complex including such phenomena as dynamic effects, random distribution of damping techniques over the actual trucks, multiple non-linear visco-elastic links in mechanical description of a given truck. For all these reasons, a system of monitoring bridges has been preferred to an unrigorous modelling, in order to get a statistical knowledge of the traffic loads applied to the bridge over large periods. This knowledge under the form of histograms will be useful in order to evaluate extreme load effects and fatigue load effects over the lifetime of the bridge. To achieve these goals, a data acquisition system based on smart sensors extracting and classifying extrema in the traffic loads signal has been developed. At each measurement site a small microsystem is dedicated to the tasks of signal conditioning and sampling, calculation and communication. Each smart sensor can communicate through a numerical data link with its neighbors or with a PC based system controller. In this paper an outline of the problem, the proposed solution based on the smart sensor paradigm, and the results which have been obtained are presented.