A comparison between Griffing's (1956) Method 1, Model I, and Method 3, Model I, using data from a 10 × 10 diallel of spring barley grown in eight different environments, revealed that inclusion of the parental generation (Method 1) caused considerable upward bias in the estimates of both the general and specific combining ability variances, whereas its exclusion (Method 3) from the analysis resulted in similar GCA∶SCA ratios in all the environments. This investigation supports the exclusion of the parental generation in combining ability analyses. The proportion of general combining ability variance was much higher than that of the specific combining ability variance although both were statistically significant. The interaction of the former with environment was much higher than that of the latter. Except for combining ability variances, the estimates of GCA and SCA effects were found to be alike in both Method 1 and Method 3. Mean performance was correlated with the general combining ability effects. Variety Cb 824 proved to be the best general combiner for earliness; it could be usefully exploited in future breeding programmes to induce earliness, as this variety possesses mostly dominant genes for earliness. Early varieties such as Otis, Bonus M-12a and Line 7-2 were also good combiners for earliness. Crosses between these varieties generally showed high specific combining ability effects.