Highly-efficient solar cells containing lead halide perovskites are expected to revolutionize sustainable energy production in the coming years. Combining these next-generation solar panels with agriculture, can optimize land-use, but brings new risks in case of leakage into the soil. Perovskites are generally assumed to be toxic because of the lead (Pb), but experimental evidence to support this prediction is scarce. We used Arabidopsis thaliana to test the toxicity of the lead-based perovskite MAPbI3 (MA = CH3NH3) and several of its precursors in plants. Our results show that MAPbI3 severely hampers plant growth at concentrations above 5 microM. Surprisingly, we find that the precursors MAI is equally toxic, while lead-based precursors without iodide are only toxic above 500 microM. These observations reveal that perovskite toxicity at low concentrations is caused by iodide ions specifically, and contrast the widespread idea that lead is the most harmful component. We calculate that iodide toxicity thresholds are likely to reach in the soil upon perovskite leakage, but much less so for lead toxicity thresholds. Hence, this work stresses the importance to further understand and predict harmful effects of iodide-containing perovskites in the environment.