Downstream flood in a large basin is generally caused by the superposition of upstream floods from different tributaries. There is certain correlation between the flood from upstream tributaries. To investigate the impact of the correlation between upstream flood on the frequency distribution characteristics of downstream flood, the frequency distribution of downstream flood was derived using the law of total probability based on the construction of conditional distributions of floods for both upstream and downstream conditions, as well as the joint probability distribution of flood variables in upstream tributaries. This study took the Fuhe River Basin in Jiangxi Province as the study area, and derived the flood frequency distribution at the Lijiadu hydrological station downstream from the Fuhe River using the joint probability distribution of flood variables from the Liaojiawan and Loujiacun stations in the upstream tributaries. The influence of the correlation of upstream flood variables on the downstream flood distribution was investigated. The results showed that the proposed method has a good fit for the flood frequency distribution at the Lijiadu Station. The correlation of the upstream flood variables was found to be an important factor determining the downstream flood frequency distribution. Specifically, the strength of correlation and tail correlation had no impact on the mean of flood frequency distribution, but had significant impact on higher-order statistical features such as skewness coefficient (C_v) and kurtosis coefficient (C_s). The C_v value of the downstream flood frequency distribution increased with the enhancement of the spatial correlation of upstream flood variables. A stronger upper-tail correlation resulted in larger values of both C_v and C_s. The increase in the strength of the correlation or upper-tail correlation of upstream flood variables led to a high flood risk of the basin. Extreme hydrological designs were more sensitive to changes in the correlation of upstream flood variables.