Objectives were to use meta-analytic methods to determine the effects of changes in DCAD prepartum on productive performance and health of dairy cows. The literature was systematically reviewed and 42 experiments that manipulated the mineral composition prepartum, including 134 treatment means and 1,803 cows, were included in the meta-analysis. Five experiments with 15 treatment means reported responses for 151 nulliparous cows. Data collected included mineral composition of diets, parity, breed, days fed prepartum diets, and means and respective SEM for urine pH, DMI, BW, BCS, productive performance, concentrations of minerals and metabolites in blood, and incidence of diseases. Random effects meta-analyses were conducted weighting by the inverse of SEM squared to account for the precision of each experiment. Models include the effects of DCAD, parity group prepartum, and other covariates that showed significance in univariable analysis. Final models were selected based on parsimony, biological relevance, and model fit. Reducing the DCAD reduced DMI prepartum but improved postpartum intake. Interactions between DCAD and parity group were observed for yields of milk, FCM, fat, and protein because reducing the DCAD improved those responses in parous cows; however, reducing the DCAD either had no effect on yields of milk and protein, or reduced the yield of FCM and fat in nulliparous cows. The model predicted that reducing the DCAD from +200 to −100 mEq/kg would increase DMI postpartum 1.0 kg/d and milk yield 1.7 kg/d. Concentrations of Ca in blood on the day of calving and postpartum increased with a reduction in DCAD, which explained the reduction in risk of milk fever observed in parous cows fed diets with negative DCAD. As the DCAD decreased, the risk of retained placenta and metritis also decreased, resulting in fewer disease events per cow in both nulliparous and parous cows. These results support the recommendation that parous cows should be fed diets with negative DCAD, but data available did not allow for detection of the ideal negative DCAD that optimizes production and minimizes health problems in parous cows.