Many hard coatings have been developed in recent years to help to solve some of the industrial problems encountered in rubbing parts in machinery. Unfortunately, coating developments for tribological applications have progressed very much more slowly than expected; each application has to be studied individually and very few guidelines are found at the design level. This paper attempts to explain why coating penetration is so slow and what can be done at the design level. It shows that the information required by the material scientist to qualify a coating (deposition method ( e.g. chemical vapour deposition or physical vapour deposition), composition, thickness, hardness, coating-to-substrate adhesion, applications wherre the coating did well, and friction and wear data produced on a pin-and-disc machine) is very different from the data needed by the mechanical engineer (Young's modulus, Poisson's ratio, coefficient of thermal expansion, thermal conductivity, density and specific heat) to design coated machine elements. The paper goes on to slow that, under well-defined conditions, hard coatings can either increase or decrease the stresses in contacts, loaded either thermally or mechanically, and that the situation is too complex to rely on intuition alone to predict whether a coating will improve or impair a given contact condition. The paper concludes by saying that hard coatings will only be used industrially if coating specialists, designers and thin film property specialists learn to work together in order (1) to develop, control and qualify coatings, (2) to identify working conditions (load, speed, temperature, etc.) and to set up models of layered systems and (3) to produce the mechanical and thermal properties of both coatings and substrates. Hard coating development is a pluridisciplinary project.