This project aims at developing and evaluating statistical methods that may provide better predictions than present methods when the number of trials is small. The number of trials in Swedish crop variety testing has decreased since the 1980s. Nowadays, the number of trials is so small that uncertainty in results has become a problem for the industry. We apply for funding for a PhD project aimed at a comprehensive evaluation of statistical methods for Swedish crop variety testing. Methods will be compared using cross-validation of existing data. Farmers should be given better opportunities to choose varieties that perform well locally. Breeders should not risk producing seeds of unstable varieties.