This paper addresses the problem of updating routing instructions for the transport of hazardous materials given real-time road weather and traffic information. A heuristic is proposed that explicitly considers the inherent multiobjective and dynamic nature of the problem and the need to produce updated instructions on-line. It is tested on the transportation network in the District of Columbia metropolitan region, where real-time data are received in a GIS environment. Solutions are compared to the a posteriori paths that could have been chosen if one could know future weather and traffic conditions exactly a priori.