Abstract: Designing efficient and robust engineering systems requires dealing with expensive computational models while taking into account uncertainties in parameters and surrounding conditions. Traditionally, safety factors have been used to compensate for uncertainties in a deterministic optimization setup. However, in order to design more efficient and safe systems, deterministic optimization is being replaced by optimization under uncertainty (OUU).  OUU is a two-loop process consisting of the outer-loop optimization and the inner-loop uncertainty propagation. Uncertainty propagation typically requires many evaluations of some expensive-to-evaluate high-delity simulation of the complex system, making OUU computationally prohibitive. In order to make OUU computationally feasible, multifidelity and multi-information source methods are becoming a necessity. I will present ways of reusing the rich source of existing information from past optimization iterations as an extra information source. This talk will deal with reliability-based design optimization (RBDO), which is a particular formulation of OUU with reliability constraints. I will describe two methods of reusing data from optimization history as an extra information source in the context of RBDO with (1) importance sampling, and (2) multifidelity active learning using Gaussian process surrogates.