Two‐dimensional (2-D) filters are used in geophysics for processing seismic, potential field, and remotely sensed data, and in other applied sciences for purposes such as image processing. These filters can be designed using optimum one‐dimensional (1-D) algorithms via mapping techniques: the 2-D desired response is mapped into the 1-D response either in the time domain or the frequency domain and mapped back after an approximation is obtained. A coefficient mapping algorithm is presented here for designing 2-D circularly symmetric filters. The 2-D frequency‐domain coefficients are mapped to the 1-D frequency axis after being sorted according to their distances from the origin. This kind of sorting brings together the coefficients of a particular passband, reject band, or transition band of the circularly symmetric filter before the coefficients are mapped into the 1-D frequency axis. As a result, there are as many prescribed bands in the 1-D domain as in the corresponding 2-D domain, which leads to optimal approximations in the 1-D domain and, consequently, in the 2-D domain. The mapping is performed between the Nyquist regions. A Chebychev or “min‐max” algorithm has been used for 1-D approximations. Due to a complete mapping between the 2-D and the 1-D domains, the 2-D filters designed via mapping show equiripple behavior similar to that of the 1-D filters. The new mapping algorithm is suitable for designing low‐pass, high‐pass, band‐pass, and band reject filters with multiple bands. The circular symmetries of the approximated responses improve with an increased number of filter coefficients. For a 2-D filter with [Formula: see text] coefficients, [Formula: see text] and [Formula: see text] must be odd; but they need not be equal. Applications of low‐pass and high‐pass circularly symmetric filters to data from a regional gravity field survey demonstrate that these filters can effectively separate anomalies of different wavelengths when there is no spectral overlap.