This paper investigates the seasonality of extreme flood events across the United States (US) and outlines the impact of large-scale climate on its spatio-temporal variability. The analysis was conducted on 318 stream gage stations from the United States Geological Survey (USGS) Hydro-Climatic Data Network (HDCN) over a period of 68 years (1950-2017), resulting in the estimation of extreme flood counts (defined as the number of days in each month when the streamflow is greater than the 95th percentile of daily distribution) and the determination of the Markham’s seasonality index for it. 17% of the stations show a high seasonality index of 0.8 to 1 and 61% of the stations have the spring months of March-April-May as the dominant season of extreme flood events. These stations are typically located in the Midwest region of the US. The second dominant season occurs during the winter period of December-January-February and is mostly seen in the coastal regions of Northwestern US, Southern and Eastern US. Inference models with the aid of Stepwise Regression and Least Absolute Shrinkage and Selection Operator (LASSO) strategy using season-ahead predictors and Negative Binomial functional assumptions were also developed to deduce the effects of large-scale climate on the rate of extreme flood occurrence. Results show that Nino12 and Nino34 (indices for El-Nino Southern Oscillation) are influencing factors in 23% of stations. NAO (North Atlantic Oscillation) and AMO (Atlantic Multi-decadal Oscillation) are significant in 20% and 16% of stations respectively. PDO (Pacific Decadal Oscillation), IPO (Inter-decadal Pacific Oscillation) and SNAO (Summer NAO) are significant in 12% of stations. These pre-season climate indicators can be used for developing forecasts of the number of floods in the upcoming seasons, which then can be used for preparation and taking pre-emptive measures to mitigate flood damages.