Technical Abstract: In addition to biomass production, alfalfa cultivars also need to express yield stability across diverse environments. The objective of this experiment was to analyze the nature of biomass yield stability in ten commercial alfalfa cultivars by evaluating performance of individual genotypes. Biomass yield was measured in each of five environments across two years, and the yield stability computed for the overall cultivar mean performance and the mean performance of each of the genotypes comprising the cultivars using the genotype x environment variance statistic of Shukla and the superiority statistic of Lin and Binns'. The GxE variance of the cultivars was not correlated with the mean GxE variance of the genotypes comprising the cultivar. However, a strong positive correlation was observed between the superiority value of the cultivar as a whole and the mean superiority value of its genotypes. Alfalfa cultivars can be stable, as measured by the GxE variance, without being composed of stable genotypes. However, cultivars identified as superior only result if the individual genotypes are also superior. The top 10% of individual genotypes selected based on GxE variance do not include any genotypes with high yield. However, truncation based on the superiority statistic selected seven of the ten top yielding genotypes. It appears that for an applied breeding program selection based on the superiority statistic would have a greater chance of improving yield and yield stability concurrently.