Recent improvements in the performance of solid coupled wheel probes and other solid coupled ultrasonic devices [1] have opened up the possibility of their wider application. If such devices are to become general purpose NDE tools then it is important that the parameters which limit their use be understood. Experience gained at Imperial College by using such devices on real engineering structures indicated that the presence of high surface roughness and dust/dirt severely affected their performance. It is the aim of this paper to quantify these effects and gain a better understanding of their origins. This increased understanding will then allow for improved design of dry coupled devices.