cc_byPiepho, Hans‐PeterLaidig, Friedrich2025-06-062025-06-062025https://doi.org/10.1111/pbr.13240https://hohpublica.uni-hohenheim.de/handle/123456789/17681Check varieties are used in plant breeding and variety testing for a number of reasons. One important use of checks is to provide connectivity between years, which facilitates comparison among genotypes of interest that are tested in different years. When long‐term data are available, such comparisons allow an assessment of realized genetic gain (RGG). Here, we consider the question of how many check varieties are needed per cycle for a reliable assessment of RGG. We propose an approach that makes use of variance component estimates for relevant random effects in a linear mixed model and plugs them into an analysis of dummy datasets set up to represent the design options being considered. Our results show that it is useful to employ a larger number of checks and to keep the replacement rate low. Furthermore, there is intercycle information to be recovered, especially when there are few checks and replacement rates are high, so modelling the cycle main effect as random pays off.engGenetic gainLong‐term trendMaizeMeta‐analysisNetwork meta‐analysisRealized genetic gainRecovery of informationVariance componentsWheat630Article2025-05-13