Different methods of analysis, such as site-specific site response analysis or the use of ground-motion prediction equations can be adopted to account for the modification of the seismic ground-motion by the near-surface stratigraphy. Each approach is associated with a different degree of complexity and associated computational and temporal cost. This thesis identifies the main limitations of these methods as broadly employed in both academia and industry and suggests a new set of methodologies for their application. In the first part of the thesis ground-motion prediction equations and their ability to model response associated with site-specific soil layering is broadly assessed. Special emphasis is paid to the description and application of the Vs-κ_0 adjustment. This ensures that the response of a ground-motion model, particularly in the upper frequency range, is representative of the characteristics of a given site-specific shallow crustal profile. Then the effect of employing a standard deviation representative of a single site is examined. This is based on the removal of the site-ergodic assumption from the published standard deviations of the models. The effect on the hazard curves and any further implications of using this site-specific standard deviation is demonstrated by performing a Probabilistic Seismic Hazard Assessment. The second part of the thesis focuses on the main limitations and ranges of applicability of 1D site-specific site response analysis. Firstly, the Equivalent Linear approximation and the Nonlinear analysis for the constitutive modified model of Kodner and Zelasko (Matasovic and Vucetic, 1993) are tested for different magnitude-distance scenarios and strain ranges. The uncertainty in the different soil properties within site-specific site response analyses and their effect on the surface predictions is also quantified. As a result, a new set of period and soil-class dependent adjustment factors are developed which can be used as an alternative to approaches based upon randomisation of the dynamic soil properties. As part of the performed analyses, the potential bias introduced through the scaling of input motions, used in site response analysis, is addressed. Finally, the significance of using different reference depths within 1D site response analysis is considered. Consequently, through progressively more complicated parametric analyses, two new approaches, are established. These can be employed individually or in combination to select a depth for site investigation as well as a reference depth for site response analysis. Ultimately, the surface spectral ordinates obtained using site response analyses and ground-motion prediction equations are compared for an active tectonic region. Each of the previously developed methods is applied in the examined case study and the results are assessed against the traditional application of the methods. In addition, the surface predictions of each of the different methods of analysis are examined in relation to their uncertainty. This comparative analysis allows one to address the question of whether increased complexity in site-response analysis has a justifiable reward in terms of the reduction of uncertainty and also enables one to identify the most appropriate level of complexity to adopt for a given project.