Construction methods for steel-concrete structures have developed significantly over the last quarter of a century, and the number of steel-concrete bridges on road and rail networks is continually increasing. However, design methods for bridges have not evolved to the same extent as those for buildings and are not completely adapted to the limit state concept. Furthermore, the behaviour in service of steel-concrete bridges is raising questions about requirements for deck slabs (durability, cracking) and about design models applicable in this domain. The main aim of this research was to study the behaviour of steel-concrete composite bridges and to propose a model for the verification of structural safety. Objectives of the research were : measurement and numerical simulation of the influence of actions on the behaviour of steel-concrete composite bridges during construction and in service in order to analyse the effectiveness of methods for improving this behaviour, analysis of steel concrete composite bridge behaviour at the ultimate limit state and proposals for simplifying design. The research progressed in the following order : Serviceability detailed study of the behaviour of young concrete, analysis of different types of prestressing and prestressing losses over time, synthesis of the various causes of tensile stresses in deck slabs and proposa1 of design criteria as a function of structural properties and deck behaviour requirements. Structural safety critical evaluation of current design methods and possibilities for the plastification of sections, definition of limits for plastic design as a function of the type of bridge, and analysis of the influence of time-dependent effects in the concrete on the behaviour at the ultimate limit state. For the serviceability limit state, a study of the actions that create tensile stresses in deck slabs led to the determination of the principal parameters influencing the formation of transverse cracks. The restraint coefficient β, defined as the ratio of steel beam and concrete cross-sectional areas, was shown to be the key parameter for evaluating the probability of cracking with time. Methods for limiting the tensile stresses in the slab were analysed and compared as a function of the type of bridge. For the verification of structural safety, a study of the actual behaviour of steel-concrete composite sections led to the justification of using plastic design calculations in order to determine the resistance of midspan sections. Limits for the application of this approach were defined, and simplified methods for considering imposed deformations such as shrinkage and temperature effects were proposed.