In this artice are reported fundamental experimental measurements on guided waves in plates with surface roughness; the experimental data are critically compared to theoretical calculations presented in Part I. All experiments, in either immersion or contact coupling mode, are modeled by the theory developed in I that exploits the phase-screen approximation. In this theory the effect of the rough surface on the received signal, on a local scale, is assumed to be restricted to the signal phase. The comparisons between experiment and predictions show good agreement in most regimes, despite the rather simplifying approximations contained in the calculation. The model is shown to fail only when the guided wave vector is close to a branch point, that is when the guided wave phase velocity approaches the compressional or shear wavespeeds of the plate. Near these values the internal partial waves comprising the guided wave strike the surfaces at grazing incidence or are evanescent, and a simple phase-screen model cannot account for this behavior. Elsewhere in the guided wave spectrum, agreement is quite good. Of practical significance is the finding that the rough-surface damping contrast can be maximized by configuring the experimental conditions to measure just below and well above the compressional critical angle. Aluminum samples, prepared by indenting or sandblasting and independently profiled to determine rms roughness, are measured in immersion and in contact transduction, the latter with wedge couplers and line sources. The influence of the roughness in immersion experiments is strongly affected by whether the upper or lower plate surface is rough, but only in the interaction zone between specular and nonspecular reflection components.