Little is known of the actual performance of existing buildings for normally Structural Engineers do not require that their structures be tested once they are built. The wide availability of computer programs to aid Structural Engineers in design and analysis is a great advantage over previous computational tools but the very precision of computer output can give the designer a false sense of accuracy. If buildings of the future are to be safe and efficient, then an assessment of the accuracy of current analytical procedures is required. This study used some of the few published measurements of the lateral and torsional dynamic characteristics of buildings to establish accurate analytical models of the structures. These measurements, for five different buildings, consisted of data on their fundamental mode shapes and natural frequencies. Initially, estimates of these characteristics were obtained by inputting traditional evaluations of the stiffness parameters for a TABS-77 program. In general, the traditional assumptions did not result in an adequate prediction when compared with the known experimental results. Improvements were made in the analytical models by incorporating "non-structural" elements or by reducing the efficiency of certain members until the fundamental mode shapes and frequencies were matched. Implications of incorrect modelling at the design stage were investigated for both static and dynamic lateral loadings. This study shows that it is necessary to match both frequencies and mode shapes if an accurate analytical model is desired. Failure to match mode shapes can seriously affect the evaluation of loads carried by the structural elements when the building is subjected to lateral loads. Internal partitions and cladding not only add stiffness to the structure but also change the mode shape. Strong evidence is provided that these nonstructural elements do carry load and do provide stiffness. This study shows that shear lag exists in shear wall elevator cores commonly occurring in buildings and this should not be neglected. Large panels buildings apparently have significant joint rotation between panels and this should be accommodated in some manner in developing an analytical model. Considerable inaccuracies have been shown to exist in present design and practice and this study provides guidance for significantly improving present analytical modelling. / Ph. D.