In order to analyze the molecular mechanisms underlying the responses of plants to different levels of drought stress, we developed a soil matric potential (SMP)-based irrigation system that precisely controls soil moisture. Using this system, rice seedlings were grown under three different drought levels, denoted Md1, Md2 and Md3, with SMP values set to -9.8, -31.0 and -309.9 kPa, respectively. Although the Md1 treatment did not alter the visible phenotype, the Md2 treatment caused stomatal closure and shoot growth retardation (SGR). The Md3 treatment markedly induced SGR, without inhibition of photosynthesis. More severe drought (Sds) treatment, under which irrigation was terminated, resulted in the wilting of leaves and inhibition of photosynthesis. Metabolome analysis revealed the accumulation of primary sugars under Md3 and Sds and of most amino acids under Sds. The starch content was increased under Md3 and decreased under Sds. Transcriptome data showed that the expression profiles of associated genes supported the observed changes in photosynthesis and metabolites, suggesting that the time lag from SGR to inhibition of photosynthesis might lead to the accumulation of photosynthates under Md3, which can be used as osmolytes under Sds. To gain further insight into the observed SGR, transcriptome and hormonome analyses were performed in specific tissues. The results showed specific decreases in indole-3-acetic acid (IAA) and cytokinin levels in Md2-, Md3- and Sds-treated shoot bases, though the expression levels of hormone metabolism-related genes were not reflected in IAA and cytokinin contents. These observations suggest that drought stress affects the distribution or degradation of cytokinin and IAA molecules.