One hundred and ninety hybrids involving 20 broad-based yellow maize populations were evaluated together with their parents for stability in expression of their grain yield potential by means of linear regression on to an environmental index. Both linear and non-linear components were observed to be significant; the linear component, in general, being the more important. Two high yielding parents, namely variety A 3D and composite Jawahar, had general adaptability, while synthetic A 22 was suited to poor environment and variety SO 4D to good environment. Seventeen hybrids had above average yield and the majority of these exhibited general adaptability.

Combining ability analysis revealed that GCA (general combining ability) was more important than SCA (specific combining ability) in the expression of mean yield. Inheritance of linear regression was controlled predominantly by GCA whereas both GCA and SCA were equally important in the expression of non-linear fluctuations. Variety Cuba was the most desirable parent, as it transmitted to its progeny high grain yield potential and low linear response. The other good combiners for grain yield were variety SC 4D and Prolific composite but these transmitted above average linear response.

Lack of association between grain yield and linear regression of both parents and GCA effects indicated their independent mode of inheritance. It suggested, therefore, that high yield can be combined with general adaptability or responsiveness to either good or poor environment.