Wildfires play an important role in vegetation composition and structure, nutrient fluxes, human health and wealth, and are interlinked with climate change. Plants have an influence on wildfire behaviour and predicting this feedback is a high research priority. For upscaling from leaf traits to wildfire behaviour we need to know if the same leaf traits are important for the flammability of (i) individual leaves, and (ii) multiple leaves packed in fuel beds. Based on a conceptual framework, we hypothesised that fuel packing properties, through airflow limitation, would overrule the effects of individual leaf morphology and chemistry. To test this hypothesis we compared the results of two experiments, respectively addressing individual leaf flammability and monospecific fuel bed flammability of 25 perennial species from eastern Australia. Across species, fuel bed packing ratio and bulk density scaled negatively with fire spread and positively with maximum temperature and burning time. Species with “curlier” leaves, higher specific leaf area, lower tannin concentrations and lower tissue density promoted faster fire spread through fuel beds. We found that species with shorter individual leaf ignition times had a faster fire spread, shorter burning times and lower temperatures in fuel beds. Leaf traits that affect the flammability of individual leaves (e.g. specific leaf area), continue to do so even when packed in fuel beds. While previous studies have focused on either flammability of individual plant particles or fire behaviour in fuel beds, this is the first time that an overarching combination of the two approaches was made for a wide range of species. Our findings provide a better understanding of fuel bed flammability based on interspecific variation in morphological and some chemical leaf traits. This can be a first step in linking leaf traits to fire behaviour in the field.