Up until today reinforced concrete has been used in most cases for the manufacturing of bridge decks. Depending on the quality of the work carried out, defects can already occur after only a few years. These defects mostly appear in the form of corrosion of the steel reinforcement due to concrete's sensitivity to de-icing salts and water. To reduce maintenance costs, which are mainly caused by corrosion of the steel reinforcement, attempts were made to eliminate the steel reinforcement in the bridge deck. This was achieved for example by replacing the whole concrete bridge deck with an FRP1 bridge deck. FRP bridge decks, besides the advantage of the absence of steel reinforcement, exhibit the advantage of a low dead load (approx. 20% of a comparable concrete deck) combined with high strength. These properties resulted in the fact that today more than 200 bridges with FRP decks are in service worldwide. Most of them need steel or concrete main girders to bridge the required span. Despite the many bridges already in service, assessment of their load-bearing capacity or deflections still remains difficult. Some of the reasons for this this are as follows: Geometry and material properties vary considerably between different FRP bridge-deck types. The problem of the connection between main girders and bridge decks has only been partially solved. No design method exists which allows determination of the stresses and deflections of composite girders, and takes the degree of composite action of the bridge deck into account. This thesis contributes to solve these problems. Experiments with two different bridge decks were carried out in order to determine the necessary system properties (in-plane compression and shear resistance and in-plane compression and shear stiffness) for the calculation of the load-bearing behavior of steel/FRP composite girders. The method developed to determine the system properties can also be applied to other FRP bridge decks (e.g. sandwich decks). In a second step, four composite girders (two with each bridge deck) were manufactured by bonding the bridge decks onto conventional steel girders. Local failure of the bridge deck, as occurs in girders with stud or bolt connections, is therefore prevented and a clear load transfer in the joint is assured. One of the two girders, with each bridge deck system, was tested statically and the other statically and under fatigue loads. The results of the girder experiments showed that adhesive bonding is a reliable connection technique, since failure always occurred first in the bridge deck and then in the adhesive layer. The stiffness and failure load of the composite girders could be increased considerably in comparison with the pure steel girder. The determined system properties concerning in-plane shear and compression stiffness were confirmed with the girder experiments. The results of the experiments with the composite girders were compared with results of an analytical design method for concrete/wood girders adapted for steel/FRP composite girders. It was shown that the load-bearing behavior of composite girders consisting of steel main girders and adhesively- bonded FRP bridge decks can be determined with good accuracy in the linear-elastic region. Furthermore a design method was developed which allows the loadbearing capacity of the steel/FRP composite girders investigated in this thesis to be determined with very good accuracy. Subsequently a parameter study was carried out in order to verify the assumption of full composite action in the adhesively- bonded joint. This is one of the requirements for application of the developed design methods. The study showed that the assumption is applicable for different adhesives and even for thicknesses up to 50 mm. ---------------------------------------- 1 Fiber Reinforced Polymer