Many biological experiments and almost all quantitative genetic studies draw conclusions from differences between families. Such differences are caused as much by heritable as by non-heritable sources. This paper investigates the impact of seed quality on the expression of quantitative variability among 20 highly inbred lines of Nicotiana rustica using ten seed categories representing different pollination times, seed sizes, seed positions in the capsule and seed maturity. Analysis of variance revealed that while the mean performance of the inbred lines remained more or less the same across all the seed categories, their individual performances differed widely, showing a highly significant interaction between lines and seed categories. This interaction accounted for up to 12% of the total variation and was largely attributable to late pollinations and immature seed. Seed categories were also the main source of heterogeneity among the within-line variances, but only immature seed lowered the correlation between the inbred means. In general, seed quality affected the developmental characters much less than the traits measured at or after flowering, and the expression of genetic variation was not critically affected by seed size or position. Immature seed, on the other hand, generated the highest level of unreliability in the comparative performance of lines and such seeds should therefore not be used in quantitative genetic experiments under any circumstances. Further implications of seed quality effects on the interpretation of genetic components are discussed in detail.