Model building methods available in the literature are not of much use for summarizing experimental data from agricultural field. As an alternative "Systematic model building (SMB)" method is suggested which is a technique of selecting the best subset of models from all possible models where terms in the model are introduced in a systematic order. We argue that the number of models to be fitted under SMB method is still very large and introduce a new method of model building which will overcome this difficulty and whore we will have to fit just one model from a well investigated ANOVA table.