AbstractMultiscale physics is the interaction of different physical processes occurring at largely separated scales. In combustion, many elementary reactions combine to only a few, but still have separated time scales. In flames, owing to the presence of diffusion, time scales manifest themselves as length scales, i.e. thicknesses of reaction layers embedded within each other. For premixed flames there results a single velocity scale, the laminar burning velocity, which in turn defines a flame thickness and a flame time as global length and time scales, respectively. The laminar burning velocity represents the simplest microscale model to be used at a premixed combustion interface.While combustion is a multiscale process, this is not so evident for turbulence. Based on the picture of a cascade process traditional turbulent closure approximations treat turbulence as a single-scale problem. Attempts to model turbulent combustion in the same way by using methods developed for non-reacting turbulent flows therefore must fail, because they ignore the multiscale nature of combustion.There is, however, a long tradition and much progress in multiscale modeling of combustion, both on the macroscale as well as on the microscale level. Unfortunately much of that work is conceived only in its particular context, not as part of a multiscale approach. For instance, papers in the TURBULENT FLAMES Colloquium and the FIRE RESEARCH Colloquium at this and at previous Combustion Symposia often take the viewpoint of macroscale modeling only, while REACTION KINETICS and LAMINAR FLAMES concentrate on microscale aspects. What seems to be needed is a more explicit reference to the needs of models developed in the other parts of the community. Furthermore, research is needed to develop suitable definitions of the interface between macroscale and microscale models.