Technical Abstract: Programs to compute multi-trait across country evaluations (MACE) were adapted to include national genomic evaluations. Strategies to account for regional genotype sharing were compared using a simulated data set in which two groups of countries shared data within but not between regions. Methods worked well unless individual countries within a region report greatly differing progeny equivalents from the same shared data, but a modified covariance matrix reduces this problem. Gains in reliability were large if young bull genotypes were evaluated in each country or region; smaller gains would occur if access to genotypes is limited. Exchange of national evaluations for young bulls is becoming much more important as reliabilities of genomic evaluations increase.