The United States was divided into three regions, on the basis of mean annual rainfall, to decrease the variability in storm-runoff constituent loads and to improve regression relations with basin and climatic characteristics. Multiple-regression analyses, in progress, are being refined to determine the best regression models for each of the storm-runoff constituent loads in each of the three regions. These techniques, when finalized, can be used to estimate storm-runoff constituent loads for gaged and ungaged urban watersheds. The preliminary standard errors of estimate for five constituents examined to date ranged from 54 to 223 percent, and the coefficients of determination (r**2) ranged from 0. 39 to 0. 94. Total storm rainfall and total contributing drainage area appear to be the most significant independent variables in the regression models. This paper is a progress report on preliminary results of estimating storm-runoff constituent loads for ungaged watersheds.