This paper presents a low-complexity demapping method for demodulating M-ary one-dimensional non-uniform constellations. For uniform constellations, there are several algorithms for searching the closest point to the received observation. However, for non-uniform constellations these techniques are not valid. We propose a novel technique based on dividing the non-uniform constellation into lattices of identical dimensions. The value of the received symbol is modified in order to provide an entry to a look-up table where the closest point is stored. This solution provides low complexity for demodulation of the in-phase and quadrature PAMs of the non-uniform constellation. The complexity is O(2+〖log〗_2⁡√M) while the exhaustive search is O(√M). The complexity implications in terms of memory and computations and the system performance are analyzed.