In the last decade probabilistic seismic risk assessment, performance based earthquake engineering and reliability or life-cycle cost based optimal design have emerged as powerful tools to guide risk-informed decisions, especially for advanced protective systems whose cost/benefit needs to be explicitly considered for widespread adoption. The fundamental component of the analysis/design methodologies that have been proposed within this context is the evaluation of seismic risk. Accurate evaluation of this risk entails adoption of appropriate excitation, structural and performance evaluation models, quantification of the uncertainties related to these models, propagation of these uncertainties to estimate risk, and, if necessary, adoption of efficient numerical optimization algorithms for performing the risk-based design. Accurate estimation of risk within this comprehensive modeling framework typically requires use of stochastic simulation techniques, because of the complexity of the adopted numerical and probability models. In this simulation-based setting the seismic   risk-based optimal design of structures might require a large number of evaluations of the system-model response (typically established through nonlinear time-history analysis), which can render the computational demand for direct application of stochastic simulation techniques prohibitive. To alleviate such computational challenges, this research develops methodologies that exploit the merits offered by the reduced order (parsimonious) modeling of structural behavior or the integration of surrogate models in a probabilistic risk quantification framework, to efficiently perform risk assessment and support enhanced risk-informed design of seismic protective systems. This goal is achieved through numerous advances: by (i) providing a framework for the calibration/selection of parsimonious hysteretic structural models; by (ii) developing an efficient stochastic search technique within the context of a simulation-based seismic risk assessment framework to support life-cycle cost based optimal design of seismic protective devices under different design scenarios; by (iii) formulating a kriging surrogate modeling framework for seismic risk assessment when the earthquake hazard is described through stochastic ground motion models; by (iv) leveraging the remarkable computational efficiency of the latter framework to formulate a versatile approach for the multi-criteria design of seismic protective systems, adopting risk quantifications that can facilitate enhanced decision support for the various stakeholders. These advances are illustrated within this thesis in two main applications: design of fluid viscous dampers based on life-cycle cost criteria and reliability-based design of floor isolation protection systems.