存在自相关时回归系数的估计是计量经济学的一个重要内容.存在自相关时原模型已转化为自回归分布滞后模型,并且广义差分模型与自回归分布滞后模型实际上是同一模型的两种不同表示形式.本文讨论了出现自相关时的回归系数估计,指出了目前参数估计存在的问题,并改进了出现自相关时的估计方法,避免了遗漏重要解释变量而降低回归方程的拟合优度的问题.“,”Estimation on model presenting autocorrelation is an important part of basic econometrics. Original model with autocorrelation is transformed into Autoregressive Distributed Lag model, which is a different form of original model as well as Generalized Difference Model. In this paper, the estimate of coefficients for autocorrelation was discussed. The problems existing in the parameter estimation were pointed out. The improvement on estimation in the presence of autoeorrelation was given, which avoids lower goodness of fit with omitting significant variables.