For lakes in watersheds that include substantial developed land the evaluation of eutrophication is facilitated by a knowledge of changes in the lake's phosphorus (P) concentrations that have occurred since development. However, background (predevelopment) concentration data do not exist for many lakes. A method for estimating background P concentrations in lakes of the Puget Sound lowland has been developed. Using a mass balance ‘Vollenweider‐type’ model, P loadings were calculated from present‐day P concentrations measured in lake water and easily measured physical characteristics for 24 lakes in undeveloped, or insignificantly developed, watersheds. From other studies, annual P loading by bulk precipitation directly on a lake's surface was estimated to be 20 kg km−2. Loading of P from forest (undeveloped) land was derived for each lake as the difference between the calculated P loading to the lake and loading by bulk precipitation. Forest land loading to each lake was converted to the annual yield (mass per unit area) of the forested portion of the watershed. The yield of P from forest land was logarithmically related to annual runoff, and the regression equation expressing this relationship explained 73% of the sample variance. By applying that regression equation to the appropriate annual runoff data the yield of P from forest land can be estimated for any lake in the study area. Loading of P from forested watershed land, calculated from forest land yield, then can be added to direct loading by bulk precipitation to estimate background P loading for each lake. By applying the mass balance model to these calculated background loadings, background total‐P concentrations can also be calculated for all lakes in the study area that have stable thermal stratification during the summer. The standard error of estimate for calculated background loadings and concentrations averages about 25% for most lakes in the area. The method presented provides reliable estimates of background (predevelopment) P loads and concentrations for Puget Sound region lakes, using data that are presently available for most lakes in the region. These estimates can be compared to present‐day P levels in developed lakes to assess the impact that land use changes have had on lake water quality.