A method of estimating large floods using a linear storage-routing approach is presented. The differences between the proposed approach and those traditionally used are (1) that the flood producing properties of basins are represented by a linear system, (2) the storage parameters of the distributed model are determined using a volume law which, unlike other storage-routing models, accounts for the distribution of storage in natural basins, and (3) the basin outflow hydrograph is determined analytically and expressed in a succinct mathematical form. The single model parameter is estimated from observed data without direct fitting, unlike most traditionally used methods. The model was tested by showing it could reproduce observed large floods on a number of basins. This paper compares the proposed approach with a traditionally used storage routing approach using observed flood data from the Hacking River basin in New South Wales, Australia. Results confirm the usefulness of the proposed approach for estimation of large floods. ¿ 2000 American Geophysical Union