The objective of this epidemiological study was to identify risk factors for Neospora caninum-related abortions in Ontario Holstein dairy herds. A total of 88 herds, consisting of 5080 cattle, and utilizing Dairy Herd Improvement (DHI) services, were divided into three groups. Case ( n = 30) and first control ( n = 31) herds were selected from 1998 and 1999 fetal abortion submissions to the Animal Health Laboratory, University of Guelph, that were histopathologically positive or negative, respectively, for N. caninum. A second control group ( n = 27) was selected from multiple sources of herds sampled within the previous 4 years that had a low seroprevalence (<7%) to N. caninum. Between May and December 1999, all available cows on all farms, in parity one or greater, were blood sampled. The sera were then analyzed for antibody to N. caninum using a kinetic ELISA. A survey administered at the time of sampling recorded information on housing, animal species present, manure management, reproduction, biosecurity practices, wildlife observations, peri-parturient cow management, herd disease history and nutrition. Production and other herd parameters were obtained from DHI records. Logistic regression indicated that the following parameters were positively associated with a N. caninum abortion in a herd: the N. caninum herd seroprevalence (OR = 1.1) , the total number of dogs on a farm (OR = 2.8), the frequency that dogs were observed defecating in mangers (OR = 2.8), the number of horses on a farm (OR = 3.1), the observed annual rate of retained fetal membranes (OR = 1.2) and the observed annual rate of cows returning to estrus after pregnancy confirmation (OR = 1.2). Factors negatively associated were the frequency of stray cats and wild canids observed on a farm (OR = 0.4 and OR = 0.7, respectively) and the housing of heifers on loafing packs (a housing pen divided into feed manger, scrape alley and bedded pack areas, OR = 0.1).