Simple, soil-free assays that can mimic drought conditions are incredibly useful for investigating plant stress responses. Due to their ease of use, the research community often relies on polyethylene glycol (PEG), mannitol and salt treatments to simulate drought conditions in the laboratory. However, while these types of osmotic stress can create phenotypes that resemble those of drought, it remains unclear how they compare at the molecular level. Here, using transcriptomics, we demonstrate that these assays are unable to replicate drought signaling responses in the Arabidopsis root. Indeed, we found a significant number of genes that were induced by drought were in fact repressed by such treatments. Since our results question the utility of PEG, mannitol and salt, we designed a new method for simulating drought. By simply adding less water to agar, our ‘low-water agar’ assay elicits gene expression responses that compare more favorably to drought stress. Furthermore, we show our approach can be leveraged as a high-throughput assay to investigate natural variation in drought responses.