This study evaluated whether differences in milk fever incidence among Swedish dairy herds could be explained partly by differences in mineral feeding during the last weeks of gestation. A case-control study was performed on dietary risk factors for a high incidence of milk fever at the herd level using information regarding feeding and management of the dry cows collected in a written questionnaire distributed by post in spring 2008. The study was conducted from September 2004 to August 2007. Data on the diets fed to dry cows, with an emphasis on the amounts of minerals (Ca, P, Mg and K) fed 3 weeks before calving and at calving, were obtained from 30 herds belonging to the 100 Swedish herds with the highest recorded incidence of veterinary treatment for milk fever (>8.8%) and from 22 herds with no recorded milk fever treatments. Multivariable logistic regression analysis showed that a linear increase in the total amount of K and less than 26 g of Mg/day fed to dry cows 3 weeks before calving was associated with an increased risk of high milk fever incidence. A large increase in the amount of dry matter (DM) fed (>3.1 kg DM extra per day at calving compared with 3 weeks before calving) was associated with a higher incidence of milk fever, but no differences were found for Ca or P intake. Breed composition, herd average milk yield and age composition of the herd did not explain any of the observed differences between the case and the control herds. The results indicate that differences in the frequency of milk fever among herds can be associated with differences in mineral feeding of the dry cows. A high amount of K in the diet may increase the risk of milk fever linearly, whereas Mg should probably be fed at a higher level than the current Nordic recommendation to prevent milk fever.