WEST Nile virus (WNV) activity has been recognised in southern Europe for decades, but evidence of WNV circulation in Spain has only recently been found, when serological and virological monitoring in populations of wild birds (Figuerola and others 2007, Höfle and others 2008, Jimenez-Clavero and others 2008, horses (Jimenez-Clavero and others 2007) and human beings (Bofill and others 2006, Kaptoul and others 2007) was carried out. However, the lack of observed bird mortalities and the practical absence of disease in horses and human beings suggest that the circulation of the virus is essentially silent. Current evidence supports a model in which WNV persists for a long time in a sylvatic, enzootic cycle, restricted to populations of wild birds that either remain asymptomatic or in which the disease goes unnoticed, with no (or only seldom) spillover that rarely progresses to affect peridomestic habitats (Zeller and others 2004).The Guadalquivir marshes in Doñana National Park, south-west Spain, and adjoining wetlands constitute a favourable habitat for WNV circulation. Recently, the authors' group found relevant seroprevalences to WNV in wild bird populations from these wetlands, particularly in common coots (Fulica atra) others 2007, 2008). During three consecutive winters (October 2003 to February 2006, a number of WNV seroconversions were detected in individual birds from this population, supporting the existence of an enzootic cycle. Moreover, the decline of WNV seroprevalence during the period examined, along with seroreversions (reduction of virus neutralisation titres by at least fourfold or disappearance of detectable virus neutralisation titres [Thrusfield 2005]) that were detected in several individuals, indicated both a decrease in the exposure of this population to WNV and a limited duration of neutralising antibodies to WNV, at least in this bird species.In 2005, a serosurvey was conducted in 156 feral horses and 194 cattle living in the Guadalquivir marshes. The study revealed a seroprevalence of 8 per cent in horses and 0 per cent in cattle (JimenezClavero and others 2007). As the feral horse population appeared to be the target for WNV infection in the area, a follow-up study was conducted in order to evaluate the evolution of the seroprevalence to WNV in the long term. This feral horse population is not considered to be peridomestic, since it lives outdoors in the marshes, far from humanised habitats, all year round. Sera taken from these horses in the summer months of 2007 and 2008 were examined using the same virus neutralisation test in microtitre format (micro-VNT) that was employed in the first survey. The horses were identified by numbered ear tags and individual microchips. In 2005, 16 of 156 horses showed neutralising antibodies to WNV (Table 1). Two years later, in 2007, 152 horses were examined, including 10 that had been VNT-positive in 2005. Of these, four remained VNT-positive and six had reverted to VNT-negative status. The remaining 142 horses gave negative results in the VNT...