The height of the observation surface above a magnetized region primarily determines the critical dimension of the smallest inhomogeneity in magnetization that can be resolved from magnetic survey data. When a rectangular block is smaller in size than this critical dimension, it appears homogeneously magnetized in the observed magnetic field. This consideration leads to the selection of a unit rectangular block of suitable dimensions with homogeneous magnetization. The magnetized region creating the anomalous field values in the area of observation can, therefore, be broken up into several blocks having different magnetizations, each block being equal in size and uniformly magnetized.The iterative method described here assumes initially that the anomalous field values are caused by a three-dimensional (3-D) distribution of magnetized rectangular blocks. The optimum orientation of these blocks with respect to geographic north is then determined. This orientation is particularly insensitive to adjustments in the dimensions of the blocks. The top and bottom surfaces of each of the blocks in one or more layers are adjusted in a least-squares sense to minimize the difference between observed and calculated field values. A method is also described for constraining the magnetization vector of each block to lie within a specified angle of the normal or reversed direction of the geomagnetic field vector. The procedure for analysis of data can also be extended to the case of anomalies over a draped surface. At the conclusion of the iterations, a 3-D distribution of magnetization is generated to delineate the magnetized region responsible for the observed anomalous magnetic field. Examples including model and aeromagnetic data are provided to demonstrate the usefulness of a generalized multibody model for inversion of magnetic anomalies.