The number of base-flow measurements needed to meet specified accuracy goals for estimating low-flow characteristics at ungaged sites can be reduced by combining an estimate based on a regression of base-flow measurements on discharges from a nearby continuous long-term gaging station with an estimate based on a regional relationship developed from regressing the low-flow characteristics on basin characteristics. The estimates should be combined by weighting each inversely with its variance from the true value. Empirically estimated values of variance of the estimated 7- day, 10-year low flow from its true value indicate that more than six or eight base-flow measurements add" little to the confidence with which such an estimate is made.