The 1921 peak discharge at Skagit River near Concrete, Washington (U.S. Geological Survey streamflow-gaging station 12194000), was verified using peak-discharge data from the flood of October 21, 2003, the largest flood since 1921. This peak discharge is critical to determining other high discharges at the gaging station and to reliably estimating the 100-year flood, the primary design flood being used in a current flood study of the Skagit River basin. The four largest annual peak discharges of record (1897, 1909, 1917, and 1921) were used to determine the 100-year flood discharge at Skagit River near Concrete. The peak discharge on December 13, 1921, was determined by James E. Stewart of the U.S. Geological Survey using a slope-area measurement and a contracted-opening measurement. An extended stage-discharge rating curve based on the 1921 peak discharge was used to determine the peak discharges of the three other large floods. Any inaccuracy in the 1921 peak discharge also would affect the accuracies of the three other largest peak discharges. The peak discharge of the 1921 flood was recalculated using the cross sections and high-water marks surveyed after the 1921 flood in conjunction with a new estimate of the channel roughness coefficient (n value) based on an n-verification analysis of the peak discharge of the October 21, 2003, flood. The n value used by Stewart for his slope-area measurement of the 1921 flood was 0.033, and the corresponding calculated peak discharge was 240,000 cubic feet per second (ft3/s). Determination of a single definitive water-surface profile for use in the n-verification analysis was precluded because of considerable variation in elevations of surveyed high-water marks from the flood on October 21, 2003. Therefore, n values were determined for two separate water-surface profiles thought to bracket a plausible range of water-surface slopes defined by high-water marks. The n value determined using the flattest plausible slope was 0.024 and the corresponding recalculated discharge of the 1921 slope-area measurement was 266,000 ft3/s. The n value determined using the steepest plausible slope was 0.032 and the corresponding recalculated discharge of the 1921 slope-area measurement was 215,000 ft3/s. The two recalculated discharges were 10.8 percent greater than (flattest slope) and 10.4 percent less than (steepest slope) the 1921 peak discharge of 240,000 ft3/s. The 1921 peak discharge was not revised because the average of the two recalculated discharges (240,500 ft3/s) is only 0.2 percent greater than the 1921 peak discharge.