The supplementary materials include figures and tables related to the study on compounding effects in flood drivers challenging estimates of extreme river floods. Figure S1 shows hydrometeorological conditions for 3,527 catchments from 1981 to 2020. Figure S2 illustrates training samples, with an example catchment in Slovakia. Figure S3 presents climatic and local attributes of the catchments. Figure S4 describes the procedure for identifying multi-driver river floods, using two catchments as examples. Figure S5 shows main driving variables for annual maximum flood events. Figure S6 discusses the sensitivity of using different thresholds to identify main drivers. Figures S7–S9 examine the impact of compounding drivers on flood events using different thresholds. Figure S10 illustrates the importance of compounding effects in two flood samples. Figure S11 explores the sensitivity of different thresholds for main interaction effects and predictive performance. Figures S12 and S13 show the impact of flood complexity on estimating large flood magnitudes. Table S1 lists candidate values for hyperparameters of LightGBM models. The supplementary materials also include various plotting positions for flood frequency analysis, such as Weibull, Cunnane, Tukey, Beard, and unbiased methods, using maximum likelihood and moments. These materials provide detailed information to support the study's findings on the compounding effects of flood drivers and their impact on flood estimation.The supplementary materials include figures and tables related to the study on compounding effects in flood drivers challenging estimates of extreme river floods. Figure S1 shows hydrometeorological conditions for 3,527 catchments from 1981 to 2020. Figure S2 illustrates training samples, with an example catchment in Slovakia. Figure S3 presents climatic and local attributes of the catchments. Figure S4 describes the procedure for identifying multi-driver river floods, using two catchments as examples. Figure S5 shows main driving variables for annual maximum flood events. Figure S6 discusses the sensitivity of using different thresholds to identify main drivers. Figures S7–S9 examine the impact of compounding drivers on flood events using different thresholds. Figure S10 illustrates the importance of compounding effects in two flood samples. Figure S11 explores the sensitivity of different thresholds for main interaction effects and predictive performance. Figures S12 and S13 show the impact of flood complexity on estimating large flood magnitudes. Table S1 lists candidate values for hyperparameters of LightGBM models. The supplementary materials also include various plotting positions for flood frequency analysis, such as Weibull, Cunnane, Tukey, Beard, and unbiased methods, using maximum likelihood and moments. These materials provide detailed information to support the study's findings on the compounding effects of flood drivers and their impact on flood estimation.