Abstract A population-based imputation procedure was used to predict the most likely genotype of un-typed loci on low density SNP maker panels to improve data integrity before genetic association and selection studies when pedigree information is not available such as in feedlot applications. It is of practical importance to evaluate the accuracy effects of imputed genotypes. In our report, a population consisting of 2246 Angus bulls that were genotyped using both Illumina Bovine3k and Bovin50 BeadChip was used. Several scenarios with varying percentages of missing SNP genotypes under a random missing pattern were simulated. Additionally, several scenarios with varying percentages of animals genotyped using the 3k and 50k panels assuming a structured missing pattern were considered. With the random missing scenarios, SNP genotypes on the Bovine50 panel were masked at random until reaching the desired missing percentage. With the structured missing scenarios, all SNP genotypes in the Bovine50 chip were masked, with the exception of those corresponding to the Bovine3 panel. The missing rates considered in this study ranged from 70% to 94% across chromosomes. Population-based imputation software fastPHASE1.2 was used for the separate analysis of each of the 30 pairs of chromosomes in the bovine genome. The results of the imputation of the random-missing SNP genotypes were similar to previous reports and accuracy rates, defined as the percentage of correct prediction of the true missing genotypes, ranging from 68% to 97% were influenced primarily by the proportion of missing genotypes. Moreover, imputation performance using structured-missing-pattern panels was impacted by the amount of individuals in reference population and level of linkage disequilibrium (LD) on each chromosome. In order to further elucidate the potential effect of incorrect imputation on genomic selection, wrongly imputed genotypes were grouped into two groups as a function of the number of incorrectly imputed alleles.