Culling is inevitable in breeding trials. It removes individuals with undesirable traits but also introduces bias in breeding value estimation since we cannot ensure the population normality assumption. Here a stochastic simulation was used to study the impact of culling on the precision of breeding value estimation. Bivariate data were simulated for different combinations of genetic parameters such as a range of heritabilities for the traits and for a range of genetic correlation between the traits. This study has three goals: Quantify the effect of culling on the ranking order of genotypes based on their breeding values. Evaluate the potential of a using a multi-trait linear mixed model to reduce that bias introduced by culling. Last, test the efficiency of three potential trial management strategies which each retain a set percentages of culled individuals to minimise the bias in breeding value estimation. The results showed that in the case one selects a small number of individuals (less than 10%) for a trait with a high heritability the impact of culling on the bias of BLUPs is much lower compared with trait with a lower heritability or when a larger number of plant is being selected for progression. The percentage of individuals culled did not influence the ranking persistence much compared with the other variables in the simulation study. Applying a multi-trait linear mixed model for breeding value estimation can reduce bias in breeding values due to culling. This approach shows a larger improvement when the correlation between the traits is high. All three trial management strategies appear to improve selection in terms of ranking persistence. However, retaining individuals systematically proves to be the most efficient trial management strategy. This study provides practical information and advice for breeding programs, which deal with culling during the plant breeding process.