With increasing congestion in major cities the occurrence of the design earthquake at the same time as the design live load is crossing a bridge is now more likely than in the past. But little is known about the effect of live load on seismic response and this report describes an experimental and analytical project that investigates this behavior. The experimental work included shake table testing of a 0.4-scale model of a three-span, horizontally curved, steel girder bridge loaded with a series of representative trucks. The model spanned four shake tables each synchronously excited with scaled ground motions from the 1994 Northridge Earthquake. Observations from the experimental work show the presence of the live load had a beneficial effect on performance of this bridge, but this effect diminished with increasing amplitude of shaking. Parameters used to measure performance included column displacement, abutment shear force, and degree of concrete spalling in the plastic hinge zones. Results obtained from a SAP2000 analysis of a nonlinear finite element model of the bridge and trucks confirm this behavior, that live load reduces the dynamic response of the bridge. The most likely explanation for this phenomenon is that the trucks act as a set of nonlinear tuned mass dampers, which are known to be effective at controlling wind vibrations in buildings. Preliminary parameter studies have also been conducted and show the above beneficial effect is generally true for other earthquake ground motions, and vehicles with different dynamic properties. Exceptions exist, but adverse effects are usually within 10% of the no-live load case.