Damage detection techniques are one of the main tools in health monitoring of structures. This paper addresses the effect of noise in the measured data on a robust damage detection method, namely statistical subspace-based damage detection technique. In this method the need of evaluating the modal parameters of the structure is circumvented which makes this method capable in real-time monitoring of structures. Moreover, this method identifies the changes in the eigen-structure of the model which makes it a robust approach to function with high amount of noise in the input data. In order to investigate the effect of noise on this method, a bridge structure located in Reibersdorf, Austria, is considered. This structure is modeled and calibrated to the real test data; subsequently the damage is modeled in one of the elements for different damage ratios. With using white noise excitation, ambient vibration test data is simulated and different noise ratios are applied to the data. A reference state of the structure is evaluated using this technique. A subspace-based residual between the reference and possibly damaged states is defined independently from the input excitations employing a ߯ ଶ test and then is compared to a threshold corresponding to the reference state. Subsequently, the effect of noise ratios in the reference state and the measured data and their difference is investigated. It was concluded that the SSDD technique is capable of estimating the damage in almost all damage ratios and even for high noise ratios in the data. Moreover, it was observed that the noise ratio difference in the reference state and measured data may be interpreted as damage, since it is reflected in the computed residual. An optimum range of the noise in the data is also assessed and proposed.