We explore the climatic factors underlying major rain induced floods in the USA. The initial hypothesis of our research was that large scale atmospheric moisture fluxes rather than local convection is responsible for annual maximum floods with exceedance probability below some threshold, given that the associated drainage basin area is larger than some threshold. This hypothesis was explored using annual maximum flood data from the USGS HCDN data set that is purportedly minimally affected by regulation and diversion of flows. Stations from large and small drainage areas were selected from 4 climate regions in the country, and in each case atmospheric moisture fluxes were computed for the major floods at each station for the 5 days preceding and the 2 days succeeding the flood event. The NCEP-reanalysis data set was used for these analyses. Composites of daily atmospheric moisture fluxes for the major flood events and corresponding daily and seasonal SST fields (from NCEP re-analysis) were developed. The initial conclusions are that floods larger than the 10 year flood in any region in the country, and for all the drainage areas investigated are determined by large scale, organized atmospheric moisture transport, which is in turn related to specific SST configurations. Thus, a case can be made for a physically based approach for the analysis of major US floods.