A new system of Ca and P requirements for lactating dairy cows was developed from meta-analysis of 53 peer-reviewed papers, with 72 minerals balance trials and 276 treatments means. The data sets represented a wide range of dietary (258 to 1000 g forage/kg DM; 121 to 258 g CP/kg DM; 260 to 496 g NDF/kg DM; 4.0 to 22.5 g Ca/kg DM; and 1.5 to 6.7 g P/kg DM) and animal characteristics (316 to 754 kg BW; 4.5 to 52.8 kg milk/d; and 8.7 to 29.0 kg DMI/d) of 3 genetic groups (66.8% Holstein; 17.7% Jersey; and other breeds 14.6%). Endogenous fecal excretion (EFE) and endogenous urinary excretion (EUE) were estimated as intercept of the linear mixed equation between mineral intake and excretion, using study as random effect. We proposed the use of the metabolizable coefficient (MC) concept to calculate dietary requirement, that represents the proportion of mineral intake that is not excreted in feces and urine, discounting endogenous excretions. Genetic group affected (P < 0.05) Ca and P concentration in milk (mean ± 95% confidence interval): Holstein = 1.17 ± 0.07 g Ca/kg and 0.89 ± 0.02 g P/kg; Jersey = 1.38 ± 0.10 g Ca/kg and 1.00 ± 0.04 g P/kg; other breeds = 1.08 ± 0.07 g Ca/kg and 0.87 ± 0.05 g P/kg. Genetic group did not affect (P > 0.05) fecal or urinary Ca and P excretions per kg DMI. Therefore, we fitted overall equations: Ca fecal (g/kg DMI) = 1.83 (±0.59; P < 0.01) + 0.032 (±0.032; P < 0.01) × Ca intake (g/d), R² = 0.66, n = 97; Ca urinary (g/kg DMI) = 0.13 (±0.03; P < 0.01) + 0.0001 (±0.009; P = 0.68) × Ca intake (g/d), R² = 0.33, n = 81; P fecal (g/kg DM) = 0.93 (±0.21; P < 0.01) + 0.016 (±0.003; P < 0.01) × P intake (g/d), R² = 0.47, n = 211; P urinary (g/kg DM) = 0.01 (±0.007; P < 0.01) + 0.0003 (±0.000; P < 0.01) × P intake (g/d), R² = 0.10, n = 126. Endogenous fecal excretions of Ca and P are estimated in 1.83 ± 0.59 g/kg DMI and 0.93 ± 0.21 g/kg DMI, and EUE in 0.13 ± 0.03 g/kg DMI and 0.01 ± 0.007 g/kg DMI. Genetic group did not affect P-MC, but Holstein had lower (P < 0.05) Ca-MC than Jersey (0.74 ± 0.03 versus 0.82 ± 0.01). Our new system reduces Ca and P dietary recommendations to non-pregnant dairy cows (10 to 50 kg milk/d) in 29% and 15% compared with the National Research Council dairy model (2001).