Abstract Weeds continue to evolve resistance to all the known modes of herbicidal action, but no herbicide with a new target site has been commercialized in nearly 20 years. The so‐called ‘new chemistries’ are simply molecules belonging to new chemical classes that have the same mechanisms of action as older herbicides (e.g. the protoporphyrinogen‐oxidase‐inhibiting pyrimidinedione saflufenacil or the very‐long‐chain fatty acid elongase targeting sulfonylisoxazoline herbicide pyroxasulfone). Therefore, the number of tools to manage weeds, and in particular those that can control herbicide‐resistant weeds, is diminishing rapidly. There is an imminent need for truly innovative classes of herbicides that explore chemical spaces and interact with target sites not previously exploited by older active ingredients. This review proposes a rationale for a natural‐products‐centered approach to herbicide discovery that capitalizes on the structural diversity and ingenuity afforded by these biologically active compounds. The natural process of extended‐throughput screening (high number of compounds tested on many potential target sites over long periods of times) that has shaped the evolution of natural products tends to generate molecules tailored to interact with specific target sites. As this review shows, there is generally little overlap between the mode of action of natural and synthetic phytotoxins, and more emphasis should be placed on applying methods that have proved beneficial to the pharmaceutical industry to solve problems in the agrochemical industry. Published 2012 by John Wiley & Sons, Ltd.