A multistate approach was used to update methods for estimating the magnitude and frequency of floods in rural, ungaged basins in South Carolina, Georgia, and North Carolina that are not substantially affected by regulation, tidal fluctuations, or urban development. Annual peak-flow data through September 2006 were analyzed for 943 streamgaging stations having 10 or more years of data on rural streams in South Carolina, Georgia, North Carolina, and adjacent parts of Alabama, Florida, Tennessee, and Virginia. Flood-frequency estimates were computed for the 943 stations by fitting the logarithms of annual peak flows for each station to a Pearson Type III distribution. As part of the computation of flood-frequency estimates for the stations, a new value for the generalized skew coefficient was developed using a Bayesian generalized least-squares regression model. Additionally, basin characteristics for these stations were computed by using a geographical information system and automated computer algorithms. Exploratory regression analyses using ordinary least-squares regression completed on the initial database of 943 gaged stations resulted in defining five hydrologic regions for South Carolina, Georgia, and North Carolina. Stations with drainage areas less than 1 square mile were removed from the database, and a procedure to examine for basin redundancy (based on drainage area and periods of record) also resulted in the removal of some stations from the regression database. Regional regression analysis, using generalized least-squares regression, was used to develop a set of predictive equations for estimating the 50-, 20-, 10-, 4-, 2-, 1-, 0.5-, and 0.2-percent chance exceedance flows for rural ungaged basins in Georgia, South Carolina, and North Carolina. Flood-frequency estimates and basin characteristics for 828 streamgaging stations were combined to form the final database used in the regional regression analysis. The final predictive equations are all functions of drainage area and percentage of the drainage basin within each hydrologic region. Average errors of prediction for these regression equations range from 34.0 to 47.7 percent. Peak-flow records at 25 regulated stations were assessed to determine if a flood-frequency analysis was appropriate. Based on those assessments, flood-frequency estimates are provided for three regulated stations. Annual peak-flow data are provided for the regulated stations in an appendix.