The meteorological conditions associated with elevated and extreme long- and short-timescale forest fire risk are investigated by validating and diagnosing the Canadian Fire Weather Index (FWI) in the context of Tuscany in Italy, and Thessaloniki, Athens and Heraklion in Greece. The aim is to provide information to assist diagnosing experiments that use output from climate models to calculate FWI values. Links are made from fire risk to the widely used FWI, and then to the underlying meteorology, complementing other more complex fire risk model studies. First, the information about Mediterranean fire risk provided by the FWI is assessed by comparing the observed number of fires per day with FWI values based on the locally observed meteorology. This shows that the FWI provides some relatively consistent information for different locations, and suggests useful FWI thresholds indicating elevated and extreme fire risk. Then, the FWI system is split according to contributions from long- and short-timescale components, in a different way than usually adopted in the literature. Using the FWI thresholds suggested above, the long- and short-timescale meteorological conditions causing elevated and extreme FWI values are diagnosed. The results may help studies that investigate what aspects of projected climate change drive changes in fire weather risk, compare fire risk calculations from different climate models, or assess how climate models can be improved to provide better fire risk projections.