The aim of this study was to compare different methods for analyzing data from a field breeding experiment with check plots where soil variability was controlled with the use of statistical and geostatistical methods. The analyzed data were obtained from field experiments investigating the yield of camelina Camelina sativa L. Crantz and crambe Crambe hispanica ssp. abyssinica Hochst. ex R.E. Fr. Prina seeds in the temperate climate of north-eastern Poland. Analysis of variance models with a completely randomized design, a randomized block design, and analysis of covariance were compared. In the vast majority of cases, the values of statistical information criteria demonstrated that the best model was the analysis of covariance where the theoretical seed yield from the check plot, interpolated with a geostatistical method, was the covariate. The results of this study can be useful for testing new cultivars in large-scale commercial farms with variable soil conditions to identify the optimal genotype for the local environmental and climatic conditions.