The most accurate estimates of median streamflows at perennial stream sites in Hawaii are those made at streamflow-gaging stations. Two alternative methods for estimating median streamflows at ungaged sites are described in this report. Multiple-regression equations were developed for estimating median streamflows at ungaged, unregulated, perennial stream sites. The equations relate combinations of drainage area, mean altitude of the main stream channel, and mean annual precipitation to median streamflow. Streamflow data from 56 long-term continuous-record gaging stations were used in the analysis. Median-streamflow data for all 56 sites were adjusted using record-extension techniques to reflect base period (1912 through 1986) conditions. Hawaii was subdivided into two geographic groups and multiple-regression equations were developed for each. The standard error of predication for the equation developed for the first group, the islands of Oahu, Molokai, and Hawaii, is 41 percent. The standard error of predication for the equation developed for the second group, the islands of Kauai and Maui, is 54 percent. A method for estimating median-streamflow, based on discharge measurements and data from nearby streamflow-gaging stations, was also developed for 27 regulated, perennial windward Oahu sites. Standard errors of prediction for 23 of the sites range from 5 to 34 percent. Median-streamflow estimates for the four remaining sites were considered poor and no measures of accuracy are provided. Discharge measurements can be used to make estimates of median streamflows at ungaged, regulated sites where the regression equations developed in this report are not applicable. Discharge measurements can also be used to make estimates of median streamflows at ungaged, unregulated sites. Estimates of median streamflows based on discharge measurements have greater standard errors than estimates based on continuous streamflow records and in general have smaller standard errors than estimates based on regression equations.