The paper presents in detail a method for approximating the spatial position of neural spike activity from tetrode recordings. The method uses a nonlinear mapping of a set of tetrode tip amplitudes into three-dimensional (3D) space, followed by a self-organizing map clustering technique. Viewed as a spike sorting method, it performs better than tetrode peak amplitudes and it is roughly equivalent with amplitude ratios. The technique's appeal to physical location may be of advantage in many investigations.