Summary 1. Bluetongue virus (BTV), a pathogen of ruminants transmitted by Culicoides midges, has emerged dramatically across Europe since 1998. Surveillance of Culicoides is currently carried out in every European country affected by BTV, most commonly using standardized light-suction traps that sample the population of adult Culicoides present. To date, however, it is not clear whether these trap catches accurately reflect the biting population of Culicoides . 2. In our study, we carried out 192 drop-trap catches at dusk on Poll Dorset sheep, drawing comparisons with surveillance samples taken using standard trapping protocol at the same site. A multiplex polymerase chain reaction technique was optimized for high-throughput processing of cryptic species to allow Culicoides collected to be identified to species level, the first time this has been achieved in this region for an ecological study. A mathematical model was then constructed to describe rates of midge biting under the meteorological conditions recorded during trials. 3. Light-trapping surveillance was found to substantially underestimate the numbers of Culicoides chiopterus present on sheep. This finding is extremely significant given that, due to the low numbers of this species caught at light traps across northern Europe, C. chiopterus had not been seriously considered as a potential vector of BTV. 4. The models constructed of successful blood feeding predict that while biting rates on sheep are significantly reduced under conditions adverse to midge flight (wind speeds that exceeded 3 mps, wind turbulence of greater than 40 degrees change in direction during the trial and solar intensity exceeding 200 Wm ‐2 ), low levels of biting can also occur under sub-optimal meteorological conditions, that have the potential to span entire days in the later part of the adult season. 5. Synthesis and applications. It is clear from these results that light-trapping surveillance does not provide an accurate reflection of the biting population of Culicoides present. Given that the surveillance systems are already in wide employment across Europe, but provide misleading results as we have demonstrated, and considering the huge economic and animal welfare impact of BTV incursions, it is vital that alternative/additional methods of surveillance are explored. Additionally, the vector competence of C. chiopterus as a potential vector of arboviruses requires urgent assessment, along with a clearer understanding of diel periodicity of Culicoides attacks throughout the adult season.