During the recent Mediterranean epizootic of bluetongue, an extensive programme of serological and vector (Culicoides biting midges (Diptera: Ceratopogonidae)) surveillance was carried out across Sicily. This paper presents the analysis of 911 light trap catches collected at the times of peak Culicoides abundance (summer to autumn 2000-2002) in 269 sites, in order to produce detailed maps of the spatial distribution of the main European vector, Culicoides imicola Kieffer and that of potential novel vectors. Whereas C. imicola was found at only 12% of sites, potential novel vectors, Culicoides obsoletus group Meigen, Culicoides pulicaris Linnaeus and Culicoides newsteadi Austen were present at over 50% of sites. However, the spatial distribution of C. imicola showed the closest correspondence to that of the 2000 and 2001 bluetongue (BT) outbreaks and its presence and abundance were significant predictors of the probability of an outbreak, suggesting that it was the main vector during these years. Although C. imicola may have played a role in transmission in several sites near Paternó, it was absent from the majority of sites at which outbreaks occurred in 2002 and from all sites in the province of Messina. All three potential novel vectors were widespread across sites at which outbreaks occurred during 2002. Of these, C. newsteadi was an unlikely candidate, as it was significantly less prevalent in outbreak vs. non-outbreak sites in Messina. It is hypothesized that the yearly distribution and intensity of outbreaks is directly attributable to the distribution and abundance of the vectors involved in transmission during each year. When C. imicola operated as the main vector in 2000 and 2001, outbreaks were few in number and were restricted to coastal regions due to low abundance and prevalence of this species. In 2002, it is hypothesized that BTV transmission was handed over to more prevalent and abundant novel vector species, leading to numerous and widespread outbreaks and probably to overwintering of the virus between 2001 and 2002. Based on catch ranges in outbreak vs. non-outbreak sites, it is tentatively suggested that nightly catches of 400 or more C. obsoletus and 150 or more C. pulicaris allow BTV transmission at a site, and provide a strategy for a fuller examination of the relationship between BTV transmission and the abundance and distribution of different vector species.