A cross-sectional serological survey for Neospora caninum was carried out on beef and dairy cattle in southern and northern Italy. A total of 111 herds and 1140 animals were tested using an ELISA assay (CHEKIT-Neospora) to detect anti-N. caninum antibodies. Management and individual data were collected and analysed both by linear and logistic multiple-regression models in order to find good predictors of the cattle seroprevalence and anti-N. caninum antibody level. At least one seropositive animal was found in 49 herds (44.1%), of which 31 (39.7%) from southern Italy and 18 (54.5%) from northern Italy. A total of 126 head of cattle (11%) were found to be seropositive and the seroprevalence was lower in southern (8.7%) than in northern Italy (16%). One of the best predictors of neosporosis seroprevalence in this study was the practice of self-rearing replacement heifers. Further risk factors were linked to higher stocking density, i.e. animals farmed in large herds and with no summer or permanent grazing practices were more likely to be seropositive than others. Farms with two or more dogs had higher herd seropositivity than farms with one or no dogs and this factor interacted significantly with the farm size and presence of poultry. Among individual characteristics, seropositivity was higher in animals sampled in mid- or late-pregnancy compared to animals either in early pregnancy or not pregnant. There was a significant interaction between the factors for pregnancy status and grazing practices. None of the epidemiological data recorded was a good predictor of the anti-N. caninum antibody level.