The limitations inherent in the seismic method of mapping faults are listed and classified. These limitations are divided into instrumental, physical, and geological. An attempt is made to treat them quantitatively and to find places where the accuracy of the seismic method can, or cannot, be improved in the foreseeable future. Attention is called to the fundamental assumptions of seismic prospecting, the prevalence in practice of contradictions of these, the inability of present methods to do what is expected of them, and the difficulty of checking their results. The role of non-structural prospecting is considered, on the basic of specific failures in current methods. A general equation for dealing with horizontal velocity changes is suggested in an appendix.