Lint yield potential, maturity, and stability of performance are important varietal traits to Tennessee cotton producers. Until recently, the stability of yield over years and locations has not been reported in performance trial reports. The objective of this study was to evaluate nineteen cotton varieties for yield and stability over nine environments that ranged in mean productivity from 371 to 1478 lb lint/acre. The variety yield trials were grown on the West Tennessee Experiment Station near Jackson, on the Milan Experiment Station near Milan, and on the Ames Plantation near Grand Junction during 1993, 1994, and 1995. The MATMODEL™ program was used to run an AMMI analysis on lint yield and the STABLE program was used to calculate the stability variance (s_i²) and yield-stability (YS_i) statistics. The AMMI1 biplot indicated that the GE interaction was related to varietal maturity and environmental season length. The two latest varieties (ST887 and SG501) had positive PCA1 scores and the four earliest varieties (ST132, H1215, H1244, and H1220) had negative scores. Only three of the nine environments had positive scores. The four earliest varieties were also the highest yielding; however, the s_i² statistic was significant for each indicating that they were also unstable. H1244, H1220, ST474, and ST132 were the most unstable varieties as their s_i² was at least 10X that of DP50. SG404 and DP50 were the most stable varieties as indicated by s_i² of 3,505 and 3,914. Five varieties (H1215, SG404, H1220, SG125, and H1244) were better than DP50 in the tradeoff between yield and stability as measured by YS_i. H1215 was the most stable of the earliest, highest yielding varieties while SG404 was the most stable variety with above average yield. A TN cotton producer should balance his variety portfolio between potentially higher yielding varieties and consistently (yet possibly lower) yielding varieties as an investor would balance the ratio of stocks and bonds.