Accurate estimates of peak flows in the Big Lost River at the Idaho National Engineering and Environmental Laboratory (INEEL) are needed to assist planners and managers with evaluating possible effects of flooding on facilities at the INEEL. A large difference of 4,350 cubic feet per second (ft³/s) between two previous estimates of the magnitude of the 100-year peak flow in the Big Lost River near the western boundary of the INEEL prompted the present study. Regression models that compared annual peak flows and attenuation of annual peak flows between successive gaging stations for the same flow event were used to estimate the magnitude of the 100-year peak flow in the Big Lost River. The 100-year peak flow of 4,790 ft³/s at the Howell Ranch gaging station was used as the starting point for this analysis. This estimate was determined by using a three-parameter log-Pearson Type III distribution as outlined in “Guidelines for Determining Flood Flow Frequency” (Bulletin 17B by the Interagency Advisory Committee on Water Data). The regression models indicated that, in the reach of the Big Lost River between Howell Ranch and Mackay Reservoir, downstream peak flows are lower than upstream peak flows. Peak-flow attenuation values for this reach of the river decreased nonlinearly as the magnitude of the peak flow increased. Extrapolation of the trend resulted in an attenuation estimate of 13 percent for this reach relative to the 100-year peak flow at the Howell Ranch gaging station. In the lower reach of the Big Lost River between Mackay Reservoir and Arco, downstream peak flows are also lower than upstream peak flows. However, in contrast to the upper reach, peak-flow attenuation values decreased linearly as the magnitude of the peak flow increased. Extrapolation of the data indicated that peak-flow attenuations in this reach of the river approach zero for flows approaching the 100-year peak-flow estimate immediately upstream and downstream from Mackay Reservoir. A regression model of annual maximum daily mean flows between Arco and the INEEL diversion dam indicated that the attenuation values in this reach of the river are nearly the same for all flows of record. Extrapolation of the linear regression of these values resulted in an attenuation estimate of 10 percent. Seepage measurements made during 1951–53 also resulted in a loss estimate of approximately 10 percent. This attenuation value, combined with the values from analyses of the upstream reaches, resulted in an estimate of the 100-year peak flow for the Big Lost River immediately upstream from the INEEL diversion dam of 3,750 ft³/s; upper and lower 95-percent confidence limits were 6,250 ft³/s and 1,300 ft³/s, respectively. Localized rainfall, even of high intensity, is not likely to produce large peak flows at the INEEL because of high loss rates (infiltration, bank storage, and channel storage) along much of the stream channel. The relatively short flow durations resulting from rainstorms historically have not provided sufficient volumes of water to satisfy local storage demands (bank and channel storage). Only after these storage demands are met do the loss rates decrease enough for significant peak flows to reach the INEEL site. An uncertain component of the present analysis is the effect of seismic activity on the 100-year peak-flow estimate. Analysis of the effect of the magnitude 7.3 Borah Peak earthquake in 1983 on normal flow conditions in the Big Lost River suggests that the joint occurrence of a large earthquake and a 100-year peak flow could significantly increase the magnitude of the peak flow at the INEEL.