Embodiments of the present invention provide methods and apparatuses for encoding and decoding a three-dimensional data point. The method for encoding the three-dimensional data point comprises: according to the position coordinates of a three-dimensional data point to be encoded, determining the maximum values of the range values of an initial block of the three-dimensional data point to be encoded in a radial distance direction, an azimuthal direction, and a height direction; performing octree division on the initial block at least once to obtain multiple first-category sub blocks; performing quadtree division and/or binary tree division on at least one of the first-category sub blocks at least once; and according to a division result of the initial block, encoding the three-dimensional data point to be encoded. The range values of the initial block constructed for dividing the distribution of spatial three-dimensional data points are different in the three directions, and therefore, in the process of division, the number of times of performing division for enabling the range values to have minimum precision is different in the three directions; when the range value has the minimum precision in one direction, then quadtree division or binary tree division is performed until all the range values in the three directions have the minimum precision, thereby reducing the number of times of division, and improving the encoding and decoding efficiency.