The comprehensive design of building systems incorporates the tasks of selection, sizing and control of devices. A simultaneous acquirement of these tasks is a necessity to achieve an overall optimal design. However, such mutual optimizations become a complex problem, implying a high computational effort A greater challenge appears once the uncertainties of boundary conditions such as weather conditions, user demands and energy costs are taken into account. A common approach to protect the suggested system configuration against the possible uncertainties is a stochastic optimization which results in a robust design. In this paper, first the sensitivity of the design to selected boundary conditions is extensively investigated. In a second step, the resulting designs of the deterministic and stochastic optimizations are compared for several uncertainties. All optimizations are setup using a previously developed framework which is extended to solve the stochastic optimization problem. The comprehensive analyses show that the achievement of a robust design is computationally demanding and not even desirable in general. However, the size of devices may vary by up to 100% when a robust design is attained. (C) 2014 Elsevier Ltd. All rights reserved.