Summary: In this critical review of the scientific literature about fire, I describe how the false notion of a link between forest fires and anthropogenic CO2 was ignited in 2006 by a fatally flawed article promoted in the science-trend-setting magazine Science, and spread like wildfire through the scientific literature and beyond, driven in part by high winds of climate modelling extravagance, while fortunately leaving large unburnt patches. There is no evidentiary basis for such a link. On the contrary, established knowledge about forest fires leads to the conclusion that dedication to teasing out such a link is preposterous: In the present circumstances starting in approximately 1900, the dominant effect is direct human impacts on land use, which causes global fire occurrences to be dramatically less than from the known long-term natural cycles (modern fire deficit). No special circumstances or regions have been correctly identified where forest fire behaviour can be attributed to CO2. Canada’s recent Fort McMurray fire is no exception. The claimed 7 g mean birth weight loss arising from mothers’ general exposure to CO2-driven southern California wildfires, like all such claims, is a product of statistical and conceptual overenthusiasm. I use concepts from the animal-behaviour scientific literature to explain how some scientists and their followers can get so carried away.