In order to solve the problems of low robustness and poor automation of current tower tilt detection methods based on light detection and ranging (LiDAR) point cloud transmission line, a tower tilt detection method based on layered minimum external rectangle and robust estimation was proposed by using the structural characteristics of tower body which was projected into standard rectangle on the level plane. Firstly, the area and proportion changes of the minimum circumscribed rectangle of each layer of point cloud were used to filter the deviation between the middle point of the rectangle and the axis point of the actual tower caused by the tower head and the high and low legs, and the center point of the minimum circumscribed rectangle of each layer of the tower body was determined as the axis point; Secondly, the elevation difference between the four sides of the minimum circumscribed rectangle and the elevation value of the point set was used to fit the axis of the tower; Finally, robust estimation was introduced into spatial line fitting to restrain the influence of other offset points on spatial line fitting by weight iteration. The experimental results show that the method has good adaptability in different tower types and different point cloud densities, and has strong ability to resist gross errors in the case of shortcomings and noise, and the deviation from the measured value is less than 0.90‰, which has strong practical application value.