Consider a TM polarized monochromatic wave that is in cident on the base of a prism from which it is totally internally reflected. If a metallic substrate is placed in the vicinity of the prism so that the evanescent fields of the reflected light penetrate the metal, and the incident angle is appropriate, a surface-plasma wave can be excited along the metal gap. Otto has analyzed this geometry and has shown both theo retically and experimentally that by measuring the reflectivity of a plane wave as a function of the angle of incidence one obtains both the speed and the range of the surface-plasma wave from the position of the minimum and the width of the Lorentzian absorption curve, respectively. In his geometry the metal thickness was much larger than the decay length of the surface-plasma field in the metal. Kretschmann con sidered a different geometry in which a thin metal film is de posited on the bottom of the prism and a plane wave incident on the base of the prism excites a surface-plasma wave on the lower side of the metal film. The excitation in this case takes place via the optical fields that penetrate the metal. As with the Otto geometry, the excitation is possible because the speed of the surface-plasma wave is slower than the speed of the light in the prism. Weber and Ford have recently shown that the fields ac companying the surface-plasma wave are larger than those associated with total internal reflection in the absence of surface-plasma waves. In their theory they assume a general coupling mechanism that is 100% efficient and equate the optical power density injected by the coupler under steady state conditions to the power dissipated by the surface-plasma wave. They show that the excitation of a surface-plasma wave at an optical frequency can give rise to an enhancement of the optical electric fields at a silver metal surface by a factor as large as 60. Their theory, however, is approximate because in calculating the enhanced fields they use the free surfaceplasma fields instead of the true fields in the presence of the excitation medium (a prism or a grating, for example.) In this Letter we wish to show theoretically that (1) it is possible to obtain an exact solution to the field enhancement Fig. 1. Three prism-coupling geometries for exciting a surfaceplasma wave: I, III, and II correspond to the Otto, Kretschmann, and our geometries, respectively.