The primary objectives of Phase I of the National Surface Water Survey were to determine the number of acidic or potentially acidic lakes and streams, their location, and their physical and chemical characteristics. To meet these objectives, a statistically designed survey was implemented. For both lakes and streams, probability samples were drawn in order to make population estimates within known confidence bounds. The National Lake Survey utilized a stratified design in which the population consisted of lakes identified on 1:250,000 maps in regions of interest. Strata were defined on the basis of region, subregion, and mapped alkalinity classes. A systematic random sample was taken within each stratum. The National Stream Survey was also stratified by region and subregion, but because a list of the extensive number of streams was not feasible, a two-stage sampling scheme utilizing a point frame was employed within strata. Both designs are flexible in that inferences about any subpopulation of interest can be made. However, because sampling intensities varied among strata, stratum-specific weights must be used when making estimates for those subpopulations that combine or cut across strata.