D.K. Roth 1985 Regression equations were developed to estimate flood peaks with selected recurrence intervals of 2 to 100 years for Ohio streams with alluvial and bedrock channels. CHannel-geometry characteristics, rather than basin characteristics, were used as independent variables. Width of active channel was the only channel-geometry characteristic significant at the 5-percent level in the estimating equations for alluvial channels. Standard errors of estimate for those equations range for the 100-year flood peak. The equations were developed from data collected at 142 gaging stations that have active-channel widths ranging from 2 to 495 feet. For streams with bedrock or firm channels, depth of the bankfull channel and active-channel width were statistically significant characteristics at the 5-percent level for all but the 2-year recurrence interval flood-peak equation, for which only active-channel width was statistically significant. Standard errors of estimate range from 33 percent for the 5-year flood peak to 40 percent for the 100-year flood peak when both significant variables are included in the equations. Standard errors of estimate range from 36 percent to 46 percent when only the active-channel width independent variable is used. These equations are based on channel-geometry data collected at 20 gaging stations that have active-channel widths ranging from 14 to 240 feet and average bankfull-channel depths ranging from 2.5 to 9.2 feet. Channel-geometry characteristics also were measured at 168 ungaged sites to provide information that can be used to better define the geographic-area boundaries in three areas of Ohio where the boundaries were previously defined in a flood magnitude and frequency report. application/pdf 10.3133/wri854175 en U.S. Geological Survey, Water Resources Division, Estimation of flood peaks from channel characteristics in Ohio reports