We propose an inverse method to estimate building and ventilation parameters from non-intrusive monitoring of heating and cooling thermal energy use of large commercial buildings. The procedure involves first deducing the loads of an ideal one-zone building from the monitored data, and then in the framework of a mechanistic macromodel, using a multistep linear regression approach to determine the regression coefficients (along with their standard errors) which can be finally translated into estimates of the physical parameters (along with the associated errors). Several different identification schemes have been evaluated using heating and cooling data generated from a detailed building simulation program for two different building geometries and building mass at two different climatic locations. A multistep identification scheme has been found to yield very accurate results, and an explanation as to why it should be so is also given. This approach has been shown to remove much of the bias introduced in multiple linear regression approach with correlated regressor variables. We have found that the parameter identification process is very accurate when daily data over an entire pear are used. Parameter identification accuracy using twelve monthly data points and daily data over three months of the year was also investigated. Identification with twelve monthly data paints seems to be fairly accurate while that using daily data over a season does not yield very good results. This latter issue needs to be investigated further because of its practical relevance.