There is a widespread consensus that drought will mostly affect present and future agriculture negatively. Generating drought-tolerant crops is thus a high priority. However complicated the underlying genetic and regulatory networks for differences in plant performance under stress are, they would be reflected in straightforward differences in primary metabolites. This is because primary metabolites such as amino acids and sugars form the building blocks of all pathways and processes for growth, development, reproduction, and environmental responses. Comparison of such differences was undertaken between the parental line and a near-isogenic line of qDTY12.1, a QTL for rice yield under drought. The comparison was informative regarding the effect of the QTL in three genetic backgrounds: donor, recipient, and improved recipient, thus illustrating the gene × gene (G × G) interactions. Such a comparison when extended to well-watered and drought conditions illustrated the gene × environment (G × E) interactions. Assessment of such G × G and G × E responses in roots, flag leaves, and spikelets added a yet more informative dimension of tissue-specific responses to drought, mediated by qDTY12.1. Data on variation in primary metabolites subjected to ANOVA, Tukey’s test, Welch’s t test, and PCA underscored the importance of the roots and demonstrated concordance between variation in metabolites and morpho-physiological responses to drought. Results suggested that for gainful insights into rice yield under drought, rather than vegetative stage drought tolerance, multiple tissues and genotypes must be assessed at the reproductive stage to avoid misleading conclusions about using particular metabolites or related genes and proteins as candidates or markers for drought tolerance.