An improved finite element model was used to obtain more accurate modeling of engineering structures. The main goal of this paper was to determine how this improved modeling procedure affects the dynamic response of buildings based on experimental modal parameters such as natural frequencies, mode shapes and damping ratios. For experimental assessment, three small, one-story, two-bay, reinforced concrete buildings with raft foundation were constructed under laboratory conditions. The initial three-dimensional finite element models, built with the ANSYS software, were used to analytically identify modal parameters, including natural frequencies and mode shapes. Following the analytical study, the ambient vibration tests were performed to obtain modal parameters, experimentally. The enhanced frequency domain decomposition method and the stochastic subspace identification method were used to identify the modal parameter. The analytical and experimental modal parameters were compared and then initial finite element models were updated to minimize the differences by changing of some uncertain parameters such as materials properties. With model updating, the maximum difference between the measured and initially obtained frequency was reduced from 28.47 to 4.88%. To show the model updating effect on dynamic response of building models, dynamic analysis with the experimental computed damping ratios and 5% damping ratio were carried out using 1992 Erzincan earthquake ground motion record. For the experimental damping ratios, the maximum differences in the displacements and stresses between the initial and updated models were obtained as 23 and 32%, respectively. These differences were calculated as 43 and 15 % for the 5 % damping ratio. Comparisons of dynamic analyses results for the initial and updated model show that the finite element model updating affects the dynamic response of the buildings considerably.