Translational genomics, the use of model species to generate knowledge about biological processes and the functions of genes, offers great promise to biotechnologists. Few studies have sought robust responses of model plants to environmental stresses, such as salinity, by altering the stress dosage or by repeating experiments in consecutive years and/or different seasons. We mined our published and unpublished data on legume salt acclimation for robust system features at the ionomic, transcriptomic and metabolomic levels. We analysed data from the model legume Lotus japonicus, obtained through six independent, long-term, non-lethal salt stress experiments which were carried out over two consecutive years. Best possible controlled greenhouse conditions were applied and two main questions asked: how reproducible are results obtained from physiologically meaningful salinity experiments, and what degree of bias may be expected if conclusions are drawn from less well-repeated sampling? A surprisingly large fraction of the transcriptional and metabolic responses to salt stress were not reproducible between experiments. A core set of robust changes was found that was shared between experiments. Many of these robust responses were qualitatively and quantitatively conserved between different accessions of the same species, indicating that the robust responses may be a sound starting point for translational genomics.