Floods affect many aspects of our lives, and our improved understanding of the processes driving the historical changes in this natural hazard can provide basic information to enhance our preparation and mitigation efforts. Here we analyze 3,885 streamgages across the conterminous United States and attribute the inter-annual variability in annual maximum discharge to precipitation and temperature. This is accomplished by first developing gamma regression models to describe the seasonal maximum discharge in terms of basin-averaged precipitation, temperature, and antecedent wetness (i.e., the basin-averaged precipitation for the season prior to the one of interest, and used as a proxy for antecedent soil moisture conditions). These seasonal models are then mixed through a Monte Carlo approach to obtain the annual maximum discharge distribution. Despite its simplicity, our results show that the developed statistical attribution approach can describe very well the inter-annual variability in annual maximum discharge across the conterminous United States.