Uniform flood frequency guidelines in the United States recommend the use of the log Pearson type 3 (LP3) distribution in flood frequency investigations. Many investigators have suggested alternate models such as the generalized extreme value (GEV) distribution as an improvement over the LP3 distribution. Using flood-flow data at 383 sites in the southwestern United States, we explore the suitability of various flood frequency models using L-moment diagrams. We also repeat the experiment performed in the original Water Resource Council report (Bulletin 17B, issued in 1982), which led to the LP3 mandate. All our evaluations consistently reveal that the LP3, GEV, and the two-and three-parameter lognormal models (LN2 and LN3) provide a good approximation to flood-flow data in this region. Other models such as the normal, Pearson, and Gumbel distributions are shown to perform poorly. Recent research indicates that regional index-flood procedures should be more accurate and more robust than the type of at-site procedures evaluated here. Nevertheless, this study reveals that index-flood procedures need not be restricted to the GEV distribution because the LN2, LN3, and LP3 distributions appear to be suitable alternatives, at least in the southwestern United States.