OBJECTIVE To identify exposures to etiologic agents and to identify characteristics that could explain risk of disease for adult cattle in herds affected by winter dysentery (WD). METHODS 229 lactating and nonlactating adult cattle (125 case and 104 control cattle) selected from 12 dairy herds. METHODS A case-control study, using multivariate conditional logistic regression and controlling for herd effects, was used to develop a model for risk factors associated with disease for each cow. RESULTS Likelihood of developing disease increased as the ELISA value for bovine coronavirus (BCV) antigen detectable in feces increased (odds ratio [OR] = 2.94 for each 0.100 increase in BCV antigen ELISA value). Pregnant cattle were less likely to develop WD, compared with nonpregnant herdmates. Cows with high acute BCV antibody titers that seroresponded had greater odds of developing disease, compared with seroresponding cows with low acute titers. However, among those cows that did not serorespond, high acute antibody titers were associated with lower odds of developing the disease. CONCLUSIONS In herds affected by WD, ill cows were more likely to shed detectable amounts of BCV antigen in their feces, and pregnancy appeared to protect cattle from the disease. The measured interaction between BCV seroresponse and acute BCV antibody titer may be evidence of an immunopathologic condition, but could also have been attributable to dynamics of the ELISA or study design. CONCLUSIONS Factors that explained a cow's risk for illness within WD-affected herds may have been surrogate measures for that cow's nonspecific and BCV-specific immune profile.