You might want to use stepwise regression when: You want to eliminate some of your less useful predictors in a complex model You have a very simple regression model and don't need the full model-building process You believe you may be violating some assumptions of least squares regression You don't have enough predictors to utilize multipl regression