A method for determining margins in conceptual-level design via probabilistic methods is described. The goal of this research is to develop a rigorous foundation for determining design margins in complex multidisciplinary systems. As an example application, the investigated method is applied to conceptual-level design of the Mars Exploration Rover (MER) cruise stage thermal control system. The method begins with identifying a set of tradable system-level parameters.Models that determine each of these tradable parameters are then created. The variables of the design are classified and assigned appropriate probability density functions. To characterize the resulting system, a Monte Carlo simulation is used. Probabilistic methods can then be used to represent uncertainties in the relevant models. Lastly, results of this simulation are combined with the risk tolerance of thermal engineers to guide in the determination of margin levels. The method is repeated until the thermal engineers are satisfied with the balance of system-level parameter values. For the thermal control example presented, margins for maximum component temperatures, dry mass, power required, schedule, and cost form the set of tradable system-level parameters. Use of this approach for the example presented yielded significant differences between the calculated design margins and the values assumed in the conceptual design of the MER cruise stage thermal control system.