Canine atopic dermatitis (CAD) is manifested by pruritus and inflammation of the skin, and most affected dogs have specific IgE against environmental allergens. Our aim was to evaluate whether in Sweden the incidence of CAD varies spatially, and to investigate possible environmental causes of such variation in a longitudinal study of CAD incidence among insured Swedish dogs. The dataset consisted of >220,000 individuals which had been covered by an insurance plan between January 1, 1995 and December 31, 2002. Out of these, 1235 CAD cases were identified and matched to postal-code areas based upon the address of the owner. Environmental risk factors we considered included averages of long-term annual rainfall, and of January and July temperatures. Initial visualization of the incidence rate of CAD (cases/dog-years at risk) expressed as empirical Bayes smoothed spatial rates indicated geographic variation. Moran's I, adjusted for population at risk, revealed significant global clustering. Both the spatial scan statistic and a local indicator of spatial autocorrelation revealed a higher incidence of CAD in the major cities. In a Poisson-regression model (with a spatial covariance structure), the incidence of CAD increased with increasing human population density, increasing average annual rainfall, living in the southern half of Sweden, and having a veterinary dermatologist in the county.