Abstract: In order to sufficiently take the influence of uncertainties on product performance into consideration,the conventional triple nested loop of reliability-based multidisciplinary design optimization(RBMDO) is decoupled by decomposition and transformed into a single recursive loop process.Accordingly,a novel RBMDO method integrating the collaborative optimization(CO) strategy and performance measurement approach(PMA) is proposed.Firstly,the CO strategy is adopted to construct the deterministic MDO model and the genetic algorithm is applied as the system-level optimizer.Then,the multidisciplinary reliability analysis and the search of most probable point(MPP) are carried out based on the PMA.Finally,if the probabilistic constraints do not meet the design requirements,the former deterministic MDO model will be reconstructed for starting a new optimization loop according to the shift strategy and the MPP information of the last iteration.The optimization loop should not stop until it converges.The optimization results of both a mathematical function and a speed reducer show that the proposed method is efficient and meaningful to engineering design.