Structural engineers routinely need to make assumptions about the dynamic properties (natural periods, mode shapes, and damping) of a building to simulate its response to dynamic loads, such as strong winds or earthquake ground motions. However, the assumed properties may significantly differ from those of the actual building, once it is constructed, due to differences between the idealized model and in situ conditions. The main objectives of this study are to evaluate how common models and assumptions used to predict the dynamic properties of buildings compare to those measured in actual buildings, and to develop improved prediction models.To this end, ambient vibration measurements were performed in 39 buildings on the island of Montréal and the dynamic properties of up to six vibration modes were identified, for each of these buildings, using the enhanced frequency domain decomposition method. Though the initial goal was to obtain a representative sample of different types of buildings, 27 of these 39 buildings turned out to be reinforced concrete buildings with shear walls providing the main resistance to lateral loads. Hence, the scope of this study was narrowed to the dynamic properties of reinforced concrete shear wall (RCSW) buildings. The measured dynamic properties of this subset of 27 buildings were then used to evaluate different models, proposed in building codes and in the literature, to predict the natural periods and damping characteristics of these types of buildings. Finally, simple models to predict the natural periods of torsion and second translation modes were proposed based on regression analyses. These models agree very well with those that have been proposed in the literature. Again, equations corresponding to different probability levels (mean, mean minus one standard deviation, and mean plus one standard deviation) were produced as a measure of uncertainty.This study should help engineers select realistic values of the dynamic properties of RCSW buildings for structural analysis and design, and should ultimately improve engineers' ability to predict the dynamic response of these buildings. Further, the proposed models could lead to improved recommendations in building codes.iii This research could not have been possible without the help of many building owners and managers, who allowed our team to perform vibration measurements in their buildings. I would also like to thank the numerous graduate and undergraduate students who were involved in these tests for both their participation and the stimulating discussions that often ensued: