Abstract: The joint distribution and the risk probabilities of three variables (flood peak, flood volume and flood duration) were analyzed to provide reference for water conservancy project planning and risk assessments. Using the flood data of Gaoyao hydrologic station at Xijiang River in Pearl River basin as an example, the primary return periods and secondary return periods of trivariate flood joint distribution and the most likely design quantiles were computed by using the asymmetric Archimedean M6 Copula and the Kendall distribution function. The main conclusions of this study can be summarized as follows: comparing the risk probabilities of trivariate flood events among the different design flood return periods, the ‘OR’ primary return periods showed higher risk probabilities and the risk probability of ‘AND’ primary return periods were lower, while the secondary return periods more accurately depicted the flood risk probability under specific design frequencies. The estimated flood design quantiles of the trivariate ‘OR’ primary return periods and three variables with the same frequency were obviously higher than that of univariate floods. The most-likely design realization of the secondary return periods can serve as the new selection for the safety and risk management of flood control projects.