In an animal model evaluation of breeding values it is assumed that the base animals are all at the same genetic level. However, in the South African Holstein population, animals of different genetic levels were imported from foreign countries, thus causing a deviation from this assumption. The effect of this deviation is considered using first lactation records from 393 458 Holstein cows. Genetic trend estimation is studied through a time trend analysis of within-Bull-yearly-Daughter Yield Deviations, or DYD`s. Bias in the estimation of trend was reduced when phantom parent groups were taken into account. The 109 385 base animals were replaced by 64 phantom parent groups. Phantom parent groups were constructed by combining year of birth, country of birth and selection intensity of the phantom parents. In recent years, foreign sires have been more affected by the exclusion of phantom parents groups in the model, than local sires, although ranking coefficients for the 15 771 sires in the analysis were in excess of 90%. Ranking coefficients for cows were also high.