The objective of this research work was to evaluate whether different stability indices of phenotypic stability vary in their repeatability. Lentil yield data of eighteen genotypes, proprietary of Dryland Agricultural Research Institute, evaluated in twelve environments over the 2002-2005 year period in four locations of Iran were used for combined analysis of variance in three datasets. I: Across locations in a single evaluation year (dataset A), II: Across locations in each of two single evaluation year (dataset B) and III: Across all of locations in three years (dataset C). Single year data of yield, of response parameters: coefficient of variation (CVi), Shukla stability variance, deviation mean squares (ER), coefficient of determination (R2), coefficient ofregression bi, Wricke ecovalence, and AMMI parameters including: SIPC!, ASV, MASV and D1 were correlated with multi year results. Among different ten stability statistics, only desirability D1 index of Annicchiarico (1997) had highly significant correlation with mean yield. CVi was significant correlated with Shukla variance and Wricke ecovalence, bi, SIPC1 parameter. Shukla variance and Wricke ecovalence indices showed highly significant rank correlation each other and also indicated significant correlated with ER, bi and ASV. Pinthus’s coefficient of determination (R2) showed significant positive correlation with ASV, MASV and bi indices. The ER statistic had positive significant correlations with the mean yield, bi, ASV and MASV stability statistics. SIPC1 parameter indicated positive significant correlation with CVi, Shukla variance, ASV and MASV. D1 Parameter had no significant correlation with ASV and MASV parameters and positive significant correlation with mean yield. Repeatability of three pair years' results (data set C) were highest for bi, R2, MASV and D1 parameters where rank correlation coefficients amounted to about 0.70. Repeatability of two pair years' results were highest for yield, SIPC1 and ASV parameters where rank correlation coefficients amounted to about 0.60. The bi, R2, MASV and D1 parameters were relatively more repeatable than SIPC1 and ASV parameters in single (dataset A) and 2-year comparisons (dataset B). Although these parameters are indices depended and proportional to yield, provides a superior way to integrate mean performance and stability into a single measure, which can be assessed visually on biplots.