Abstract: In this paper at first we introduce general linear model whioh inoludes regression, stochastio regression, autoregression, general autoregression, threshold autoregression, and mixing regression of them, and so on. General linear models have been used widely in practice. The consistent properties of least square method in fitting general linear models were developed quickly in last few years. Secondly in this paper we indroduce the main results in this development inoluding about not only parameter's estimation but also variable's selection of general linear models.