This thesis proposes a model for predicting method stability using object-oriented metrics. This model predicts possibility that a method will be affected from change at an attribute or a method of source code. Thus, suitable actions may be planned when a change occurs. This research uses 3 metric sets including method size, method complexity and coupling between methods. The stability used to construct the model is gotten from change effect when source code had been changed. The considered changes are static changes and syntactic impacts which occur at an attribute or a method of source code. This thesis also studies the relationship between stability of a class and stability of methods that are member of that class in order to study the possibility of using the method stability to explain the class stability by choosing maximum, minimum and mean of method stability to study. The result at 0.01 significant level shows that mean of method stability has highest correlation with class stability, and maximum of method stability has lowest correlation with class stability. Therefore, a mean of method stability should be used as an indicator of a class stability.