DescriptionDebonding of reinforcement in highway bridge decks can result from vibrations induced by large vehicular live loads in adjacent lanes. These detrimental effects can be evaluated by comparing rebar vibrations during concrete setting to experimentally established limits in terms of peak particle velocity. However, methods of directly measuring rebar velocity are limited. It is often more feasible to process accelerometer measurements to obtain velocity data indirectly, but common processing techniques such as direct integration will produce errors due to unknown, non-zero initial values combined with random noise on the structure. Using a combination of numerical and subjective analyses to mitigate the various sources of error, an approach is developed to estimate velocities and displacements from raw accelerometer data. Initial calibration of the algorithm is achieved by conducting a comprehensive field testing program for two independent highway bridges. The estimated dynamic response of the bridge girders compare well with the measurements taken by a Laser Doppler Vibrometer in the field. In most cases, the time histories of velocity and displacement are accurately predicted by the algorithm. The response of a bridge superstructure is monitored for an extended period, encompassing the pour of two full spans and several hours after initial concrete placement. The investigation is performed systematically; starting with the girders, progressing to the bridge deck, and culminating in the determination of rebar velocity relative to the surrounding concrete deck. Comparison with established vibration limits suggests that the vibrations during the concrete initial setting period should not pose any significant risk of debonding.