The objective of this study was to investigate causes of inflation in genomic predictions for dairy cattle. The simulated data included phenotypes, pedigrees, and genotypes, mimicking a dairy cattle population, which was selected by breeding values or not selected. With the simulated data, genomic (G)EBV were calculated with a single-step genomic BLUP and compared with true breeding values (TBV). Phenotypes and genotypes were simulated for 10 generations and the last 4 generations, respectively. Phenotypes in the last 2 generations were removed to predict breeding values using only genomic and pedigree information. For comparison, (G)EBV were also calculated using all phenotypes and genotypes in 10 generations. Pedigrees with and without inbreeding and pedigrees with unknown dams were used to construct the pedigree-based relationship matrix (A). Regression coefficients (b₁) of TBV on (G)EBV were calculated to investigate inflation in GEBV. In addition to the simulation study, inflation in GEBV for 18 linear type traits of US Holsteins were examined as well. Regression coefficients of daughter yield deviations on GEBV for young genotyped bulls were calculated. The results from the simulation indicated that GEBV for bulls were inflated regardless of selection whereas EBV were not inflated with no selection. The inflation was greater with no inbreeding or with no dams in A. On the other hand, inflation in GEBV for cows were minimal with no selection or with genotypes in all generations. For linear type traits, GEBV (and parent averages) were always inflated (b₁ < 1.0). To minimize the inflation, A and the genomic relationship matrix (G) should be consistent; including exact inbreeding in A is one way and using a weight (ω) < 1.0 on the inverse of A for genotyped animals (A₂₂⁻¹), which increases the pedigree contribution (A⁻¹ – ωA₂₂⁻¹) for genotyped animals, is another way. Smaller or current additive genetic variances could be useful. In dairy cattle, known (accurate) pedigree information and consistency between G and A (and A₂₂) could be essential to reduce the inflation in genomic predictions for young genotyped bulls.