The objective of this report is to present the various analysis techniques in a logical fashion, briefly discussing and illustrating the strengths and weaknesses of each technique with monitored data from several of the LoanSTAR buildings, and providing the reader with complete references of work previously published. As of December 1993, we are reporting retrofit savings on a monthly basis in 34 sites consisting of 50 buildings. Some of the issues addressed relate to determining conditions when statistical regression techniques are preferable to engineering models, to the different forms of single variable and multiple variable regression models and their advantages or disadvantages, to when a specific regression model is to be used and to why inspection of model residuals is important. Other issues such as the uncertainty associated with the savings estimates and the effect of building operating schedules have also been addressed in this report. The framework laid out in this report should be useful for both the practicing professional and the researcher involved or interested in energy conservation as a whole.