This paper presents a method of genotypic stability analysis for regional variety trials. Based on the principle of structural relationship analysis, the genotype‐environment interaction effect of a variety is partitioned into two components. They are the linear response to environmental effects, which is measured by a statistic α̂, and the deviation from the linear response, which is measured by another statistic λ̂ A perfectly stable variety has (α, λ)=(‐1, 1) and a variety with average stability has (α, λ) = (0, 1).Seedlings in three series of potato maincrop regional trials were analyzed for the genotypic stability of their marketable tuber yields. The results showed that the highest yielders were unstable, while the seedlings with average stability had about the same yielding ability as the check varieties.