Methods of estimating flood magnitudes for recurrence intervals of 1.5, 2, 5, 10, 25, 50, 100, 200, and 500 years have been developed for rural streams in Alabama that are not affected by regulation or urbanization. Regression relations were developed using generalized least-squares regression techniques to estimate flood magnitude and frequency on ungaged streams as a function of the basin drainage area. These methods are based on flood-frequency characteristics for 169 gaging stations in Alabama and 47 gaging stations in adjacent states having 10 or more years of record through September 2003. Graphical relations of peak flows to drainage areas are presented for sites along the Alabama, Coosa, Tallapoosa, Tennessee, Tombigbee, and Black Warrior Rivers. Equations that account for drainage area and percentage of impervious cover as independent variables also are provided for estimating flood magnitudes on ungaged urban streams (taken from a previous report).