Some of the recent floods in the United States have indicated that, in any appraisal of the potentialities of a river system for producing floods, more significance than has perhaps been customary should be attached to the magnitude of the great floods of the past, as disclosed by Nature's records of them. A conspicuous part of the work of rivers in the processes of dynamic geology is associated with the occurrence of major floods, and significant information with respect to such floods may be appropriately and reliably interpreted from the flood plains, terraces, earlier river channels, and other evidences of erosion and deposition which record a river's past behaviour.