Many steel bridges have suffered severe diaphragm (cross-frame) damage during recent earthquakes around the world. The relative role played by intermediate and end-diaphragms in providing lateral load resistance, and the consequences of diaphragm damage on bridge seismic response have not been studied. These bridges are also frequently supported by seismically vulnerable non-ductile substructures whose seismic retrofit can be, in many cases, a rather costly operation. This research addresses three related issues to the seismic behavior of slab-on-girder steel highway bridges: (i) the impact of diaphragm on their seismic performance, (ii) innovative methods for their seismic retrofit and (iii) corrosion effects on the strength and ductility of their steel members. To study the first issue, research is conducted to quantitatively investigate the impact of diaphragms on the seismic response of these steel bridges. It is shown that a small end-diaphragm stiffness is sufficient to make the entire superstructure behave as a unit in the elastic range. However, a dramatic shift in seismic behavior occurs once those end-diaphragms fail, with a sizeable period elongation, considerably larger lateral displacements and higher propensity to damage due to P-$\Lambda$ effects. For seismic retrofits of these bridges, this research consists of four essential parts: development of the ductile end-diaphragm concept, development of a design procedure, a series of inelastic analyses to validate the concept, and testing of all scale models for the proposed end-diaphragms to further verify the concept and to validate the proposed detailing. First, to take advantage of the benefits granted by the presence of a steel superstructure, an innovative, economical effective and simple to implement seismic retrofit strategy, using ductile steel bridge end-diaphragms (such as shear panels, eccentrically braced frames and triangular-plate added damping and stiffness devices) has been developed. Second, a step by step design procedure for ductile end-diaphragms is proposed and potential limits of application are indicated. In the third part, the DRAIN-2DX and ADINA programs are used to conduct nonlinear inelastic analyses of these retrofitted bridges and investigate their expected seismic behavior. Test results demonstrate the effectiveness of the proposed ductile end-diaphragms as effective passive energy dissipation systems in slab-on-girder steel bridges. For the third issue, the effect of corrosion on the strength and ductility of steel members is also studied to investigate the long-term performance of the ductile end-diaphragms proposed here. To generate preliminary data, a few rusted pieces taken from an existing steel bridge have been subjected to numerous cycles of alternating plasticity in flexure. Specimens had up to a 60% loss of cross-sectional area due to corrosion. This limited test program revealed that, while stable hysteretic behavior comparable to that of unrusted specimen is possible, premature failure under alternating plasticity can typically develop (in spite of satisfactory ductile behavior under monotonic loading). Irregularities along the severely rusted surface apparently act as crack initiators and precipitate crack propagation throughout the section. (Abstract shortened by UMI.).