Annual peak-flow data from 128 U.S. Geological Survey streamflow-gaging stations in Connecticut with at least 10 consecutive years of record were used to estimate peak-flow magnitudes for 1.5-, 2-, 10-, 25-, 50-, 100- and 500-year recurrence intervals (exceedance probabilities of 0.67, 0.50, 0.10, 0.04, 0.02, 0.01, and 0.002, respectively). Peak-flow frequency analyses of annual peak flows through the 2001 water year were performed using the procedures in the publication 'Guidelines for Determining Flood-Flow Frequency,' commonly referred to as Bulletin 17B, by the Interagency Advisory Committee on Water Data (1982). A generalized skew coefficient of 0.34, with a standard error of prediction of 0.51, was developed to improve peak-flow frequency estimates in the state; this replaces the generalized skew coefficients for Connecticut shown in Bulletin 17B.