Abstract: The performance based design for structures should be based on the theory of probability; thus establishing a reasonable probabilistic seismic demand model of structures is very important. This paper aims at the regular girder bridges of highway. 8 representative samples are built by OpenSees software and the time history analyses are carried out with the selected 2390 earthquake records at three different site conditions by cloud approach. The spectral acceleration at the fundamental period with 5% damping (S_a) is selected as the ground motion intensity measure and the displacement drift ratio at the top of pier (D) is selected as the engineering demand parameter. The probability distribution between S_a and D for different bridge samples is obtained at three different site conditions. The result shows that the EDP follows an approximate lognormal distribution with different level of IM at three different site conditions; the relevant simplified probabilistic seismic demand model is built by the analysis results. Example of verification shows that the simplified probabilistic seismic demand model has a high calculation accuracy to simplify the calculation of probabilistic seismic demand for regular bridges.