Work on reactive planning evolve from the fact that classical planners were unable to adapt to a changing environmnet or reason in limited time. Much work in robot planning during the late eighties concentrated on resolving these problems. A problem that has been given much less attention in literature are expert-system style diagnose system which face the same difficulties of adaptivity and time limitations. This paper coins the notion of reactive diagnosis to denote diagnose and explanation problems where time limitations and adaption to changes are important factors. The weighted abduction principle is suggested as a suitable approach to these problems, and an algorithm is given that exploits some of the proporties necessary for reactive diagnosis. The use of the algorithm is illustrated by an example application from interactive route guidence.