The first commercial use of synthetic pesticides for crop protection dates back to the 1940s, followed by a fast spreading of their use and the development of a large number of compounds. In contrast to synthetic pesticides that are nowadays designed with the help of artificial intelligence that includes computational science and combinatorial chemistry, natural pesticides are the results of long evolutionary processes involving specific host-pathogens, predator-prey and competitor interactions. For these reasons, natural pesticides are often more specific and less harmful to the environment. Natural pesticides are very diverse and can be found in various living organisms. In the present study, we investigated differences in the physicochemical space occupied by synthetic and natural pesticides. In this respect, we measured the mean and breadth of synthetic and natural pesticides, as well as the overlap between these groups in a reduced physicochemical space derived from a set of 44 physicochemical variables. We showed that physicochemical space strongly differs between synthetic and natural pesticides and could be determined with 93-100% certainty, a result comparable to differences observed in drugs. Importantly, the physicochemical space occupied by synthetic pesticides was 2.6 fold smaller than the one of natural pesticides and toxicity potential was lower in the latter. In conclusion, our work showed that the design of commercialized synthetic pesticides is underexploiting the possible physicochemical space of known natural pesticides, likely due to specific constraints. Such limitations should trigger the development of efficient natural pesticides avoiding as much as possible detrimental effects on non-target organisms