To study the flood coincidence law of the Lishui River and Songzi River, based on the measured mean daily flow data at the Shimen Station in the Lishui River and at the Xinjiangkou Station and Shadaoguan Station in the Songzi River from 1959 to 2020, a non-uniform GAMLSS model was used to optimize the marginal distributions of the annual maximum flood peak and 7-day continuous annual maximum flood volume. A two-dimensional joint distribution model was constructed based on the Copula function, and the probability of flood coincidence under different combinations of magnitudes was calculated. The results show that the flood sequence in the Song-Li Region has undergone significant variation, and the actual flood characteristics of the region can be described by the time-varying log-normal distribution, Gamma distribution, Weibull distribution, and Copula joint distribution. The typical annual flood encountered in the Song-Li area is mainly the Songzi River flood process and the Lishui River flood. The probability of co-occurrence of the annual maximum flood peaks or 7-day continuous maximum flood volumes of the Lishui and Songzi rivers decreases with the increase of flood magnitude, and the probability of flood coincidence in the Lishui River and the eastern branch of the Songzi Rivers is higher. When a 1 000-year flood occurs in the Lishui River, the probability of co-occurrence of the 1 000-year flood in the Lishui River and small- and medium-sized floods in the Songzi River is higher.