Phenotypic stability and genotype × environment interactions of reproductive traits, yield, and yield components were estimated in pea by growing eight genotypes (cv. Ichiban, Huka, Maro, Birte, Progreta, Combi, Greengolt, and Carpo) in nine environments (three planting densities × three sowing dates). The regression coefficient (β) of the character against its average environmental performance of all lines was used as a measure of both stability and adaptation. Yield is likely to respond slowly to direct selection as indicated by a small genotypic effect and low broad dense heritability (h2bs). Broad sense heritability of number of ovule initials per pod was extremely high and there was no G×E interaction. The genotypic component of this character was very high. Early generation selection for number of ovule initials per pod is feasible. In contrast, the number of flowers, initiated pods, and flowers aborted per plant had low h2bsand a low genotypic effect, while the number of pods aborted per plant had medium h2bsbut a low genotypic effect with high G×E interaction, suggesting poor response to selection. Ovule abortion percentage had a high genotypic effect. The G×E effect was negligible for this character and it had high broad sense heritability. Breeding this character for low ovule abortion should be promising. Broad sense heritability of pods per plant was very low and G×E interaction was high. Breeding this character for specific adaptation and for stability across environments is not feasible with these lines. The number of peas per pod showed neither high genotypic effect nor G×E interaction effect. Progress in selection for this character would be slow. The effect of genotype was predominant for 1000 seed weight which had high heritability. Improvement of yield through selection for this character is possible.