Assessment of cultivar performance in trials conducted across a range of locations and over years is often difficult because of the presence of significant location x year interaction. Methods for separation of environmental (location x year) variation into predictable and unpredictable components are available in literature. In the present paper an index, based on cultivar yield and the predictable component of environmental variance, MS(Y/L), is proposed for selecting cultivars simultaneously for high yield and stability. Two sets of rice data from All India Coordinated Rice Improvement Program, Directorate of Rice Research, Hyderabad, are used to illustrate the superior performance of the index method
