ABSTRACT. Datasets of US and Israeli Holsteins were used to evaluate the impact of older generations on ability to predict EBV of young genotyped animals in traditional and single-step genomic BLUP. Inclusion of two (A2) or all (Af) ancestor generations was also evaluated. A total of 34,506 US and 1,305 Israeli bulls were genotyped. Thresholds for data deletion were based on 5 years interval. The number of generations deleted without reduction in accuracy depended on data structure and trait. For US Holsteins, removing 3 and 4 generations of data did not reduce accuracy for final score in Af and A2, respectively. For Israeli Holsteins, the effect of removing older generations depended on the genetic origins of the bulls. Therefore, truncating older data does not decrease the accuracy of young genotyped bulls, while reducing computation requirements and helping to find problems in the population structure.