Abstract: In this paper, the estimation of the direction of the regression ooefficient of the general regression model is considered. Li and Duan use maximum likelihood type criteria to get a consistent series of estimators of the direction. The condition, which they imposed on the distribution of the regressor variable X in their construction of the estimators, is very restrictive. In this paper, to avoid the restrictive condition, a nonparametric estimator of the direction of the general regression model is constructed. The estimator is proved to be consistent.