The invention concerns a method for determining a position ϕ of a position indicator of a position measuring system, in which the sensor signals a1 to ar of a number of r sensors are recorded as a signal vector {right arrow over (a)}=(a1, . . . , ar), a measurement vector {right arrow over (q)} is formed according to {right arrow over (q)}=g({right arrow over (a)}), a 2-component vector {right arrow over (p)} is calculated according to {right arrow over (p)}=M·{right arrow over (q)}, wherein M is a 2×n matrix, and the position ϕ is determined by means of a predetermined function ƒ({right arrow over (p)}) to ϕ=ƒ({right arrow over (p)}), wherein the function ƒ({right arrow over (p)})=ƒ(p1,p2) is based on the equations p1=cos(ϕ) and p2=sin(ϕ). The matrix M is determined in a calibration phase in such a manner, that the transformation {right arrow over (p)}=M·{right arrow over (q)} is a linear mapping, which maps a number m of measurement vectors {right arrow over (q)}k on 2-component vectors {right arrow over (p)}k, wherein the tips of the vectors {right arrow over (p)}k substantially lie on a circle or on a circular arc. The method allows the elimination of external fields.