A geographic information system and K-function analysis were used to evaluate the spatial association of canine serological results for Rickettsia conorii, the causative agent of Mediterranean spotted fever (MSF), and clinical cases of MSF in humans in Piemonte, northwest Italy. The residences of dog owners were clustered in two rural villages in the province of Cuneo, where two human cases of MSF occurred in 1997 and 1998. Eighteen out of 116 dogs examined were positive by indirect immunofluorescent assay (IFA+, titre > or =1:160) for MSF. K-functions were compared for IFA+ dogs and for all dogs sampled. Monte Carlo and bootstrap simulations demonstrated that clustering of IFA+ dogs was significantly greater than clustering of all dogs, at distances of less than 0.6 km from human cases of MSF. Logistic regression analysis indicated that the risk of being IFA+ was highest for dogs residing within the first quartile of distance (0.7 km) from human cases of the disease, and for dogs that were not confined. However, year and season of blood collection were not associated with IFA status. It was concluded that a relatively high dog population density along with a rural or semi-rural environment favours the occurrence of emergent foci of MSF in the province of Cuneo.