Current knowledge suggests that there is a low-level and recurrent circulation of West Nile virus (WNV) in Europe, with sporadic human and/or equines cases. However, recent events indicate that this picture is changing, raising the possibility that Europe could experience a modification in the virus' circulation patterns. We used an existing model of WNV circulation between Southern Europe and West Africa to estimate the sample size of equivalent West Nile surveillance systems, either passive (based upon horse populations and sentinel veterinarians) or active (sentinel horses, sentinel chickens, or WNV genome detection in trapped mosquito pools). The costs and calendar day of first detection of these different surveillance systems were compared under three different epidemiological scenarios: very low level circulation, low level recurrent circulation, and epidemic situation. The passive surveillance of 1000 horses by specialized veterinarian clinics appeared to be the most cost-effective system in the current European context, and estimated median dates of first detection appeared consistent with recent field observations. Our results can be used to optimize surveillance designs for different epidemiological requirements.