To comply with the classification requirements listed in 10 CFR 61, operators of nuclear power plants are recommended to identify and quantify the concentration of seveal nuclids in low-level radioactive wastes(LLWs). Much of the specified radionuclides can not be easily measured in routine plant analyses. Many indirect methods has been suggested to determine the radionuclied concentrations upon which the waste classification is based. Such indirect methods include the use of scaling factors which infer the concentration of one radionuclide from another which can be measured easily. In this study, correlation analysis is performed to find out the important variables. Regression equations are attempted to provide a means of indirectly determining the concentraion of the difficult-to-measure nuclides based on the result of the correlation analysis. Then residual analysis and the corresponding stepwise procedure are followed to check the regression model and select the best regression equation. The regression equation whose log mean dispersion is smaller than 10 is suggested as the appropriate correlation formula. Most of the quadratic regression equations are turned out to be able to use as a correlation formula. But, TRUs show log mean dispersions which are much larger than 10. It is concluded that the mechanisms of their formation and disappearance are much more complex. And it is also difficult to select the key nuclide. In the case of TRUs, further study is required to find out the relevant correlation formula.