Design streamflows are frequently used in water quality studies to provide adequate protection against pollutant exposure periods of a given duration. By analyzing the effect that simple streamflow dilution has on x-day average exposure levels of a pollutant, it appears that the x-day harmonic mean flow is a more meaningful statistic to use in computing design flows than is the customary arithmetic mean flow. The significance of the result was examined by computing design flows for sixty rivers throughout the country. For 7- and 30-day average annual low flows the impact was minimal. Substantial differences were found for the overall mean daily flow--a design flow suggested for use with water quality criteria based on protecting human health against lifetime exposures. The overall harmonic mean daily flow is typically 20 to 60 percent as large as the arithmetic mean daily flow. For ungaged streams it can be estimated from estimates of the arithmetic mean flow and the 7-day, 10-year low flow by regional regression analysis.