The potential health impacts from smoke are well known. Protracted exposure of rural and urban populations, particularly the more susceptible groups including the elderly, people with impaired cardiovascular function and allergies, extends the risk from bushfires to the entire population and, in some cases, due to the number of people exposed, may constitute the greatest risk to health. The cost of these impacts can be substantial. Consequently smoke management is now a major issue for fire agencies.  The exposure risks vary widely with the class and location of the fire events, and range from extensive fumigation of SE Australian populations from fires that persist for weeks to localised impacts from small fires.  The FireDST project has been developing techniques for assessing the impacts and risk through a series of case studies. The key issue to emerge is not the total emission of smoke, but the extent to which the emissions mix back the surface layer, and the persistence of the smoke in the air shed. Of the three wildfire events investigated, the alpine fires of 2003 and 2006 impacted most of Victoria, including Melbourne for protracted periods, contrasting with the extreme Kilmore fire of 2009, which, in comparison, was short-lived with limited smoke impacts. The  mechanisms of smoke emission and dispersion observed in these events  has some similarities to processes observed in the regeneration burning, where injection height and boundary layer ventilation rate determine the  surface concentration and duration. The key features of these case studies that lead to their contrasting outcomes form the subject of this presentation.