The discovery of genes that can be used to increase plant tolerance to environmental stress has practical implications for agriculture, since knowledge at the molecular level can potentially be translated from model plants to crops or from tolerant to sensitive cultivars. For more than a decade, researchers have attempted to identify transcriptional and metabolic pathways involved in stress tolerance using functional genomics tools. In some cases, promising results were obtained when a clear causal link was found between transcripts and tolerance/sensitivity to stress. However, recent reports question the global translational power of functional genomics for biotechnological applications, as one of the main limitations seems to be the large variability in gene expression. Transcript-level variability under stress has not been considered of interest in the scientific literature because it is intuitively obvious, but most reports seem to naively overlook the consequences. Here, three case situations are reviewed (variability between genotypes, variability due to environmental interactions and variability between stressors) in support of the concept that inherent transcript-level variation in biological systems may limit our knowledge of environmental plant tolerance and of functional genomics in molecular stress physiology.