There are many options to define the seismic input for structural analysis. Natural recordings are considered as the most attractive option. But when insufficient - previously recorded - earthquake accelerograms are available or when they do not belong to the same seismic environment for a particular case study, it is common in practice to select other remotely recorded ground motions – or artificially generated ones - and ‘scale' them considering number of criteria to get an unbiased estimation of seismic demand. The study presented in this paper aims to investigate the efficiency of a proposed scaling scheme for ground records, which is based on matching the code design response spectrum at the fundamental period of the bridge, (Sd (T1)), or minimizing the error between this spectrum and the record-specific response spectrum at different control periods associated with the main vibration modes governing the response of the bridge. Scaling records to match Sd (T1), successfully predicts the response of the bridge (with a relatively limited variation in results from different records) even when using general ground records and not site-specific actual records. The study is accomplished by applying the proposed scaling scheme to some selected continuous bridge systems commonly encountered in Egypt. These bridges feature four equal spans with lengths of 25, 45 and 65 m investigated to represent short, medium and long span bridges, respectively. A set of twenty three international ground motion records were chosen from worldwide available strong motion database to assess the efficiency of the proposed scaling technique.