The Electrogastrogram (EGG) is a new diagnostic tool for the stomach of the human being. Its first measurement was done 75 years ago. At first, the analysis of the EGG signals was very difficult, since the pattern of this signal shows no meaning of the stomach diseases or state. In this work, we try to represent the Stomach State using the Autoregressive Moving Average (ARMA) Coefficients to be able in a later stage to discriminate between different stomach states (Hunger, Digestion, Before, and After Haemodialysis). Three Off-Line ARMA Coefficients estimators and an adaptive one are discussed. These techniques are applied to real EGG signals acquired cutaneously from the abdomen of the human being in one of the army hospitals. Their performances are tested and evaluated to elect one of them for later applications. The results show the efficient accuracy of the recommended algorithm.