Although natural product-based discovery strategies have not been as successful for herbicides as for other pesticides or pharmaceuticals, there have been some notable successes. Phosphinothricin, the biosynthetic version of glufosinate, and bialaphos are phytotoxic microbial products that have yielded commercial herbicides. Cinmethylin, a herbicidal analogue of cineole, has been sold in Europe and Asia. The triketone herbicides are derivatives of the plant-produced phytotoxin leptospermone. These products represent only a small fraction of commercialized herbicides, but they have each introduced a novel molecular target site for herbicides. Analysis of the literature reveals that phytotoxic natural products act on a large number of unexploited herbicide target sites. The pesticide industry's natural product discovery eorts have so far concentrated on microbially derived phytotoxins, primarily from non-pathogenic soil microbes, involving the screening of large numbers of exotic isolates. Plant pathogens usually produce potent phytotoxins, yet they have received relatively little attention. Even less eort has been made to discover plant-derived phytotoxins. Bioassay-directed isolation has been the preferred method of discovery after a producing organism is selected. This laborious approach often leads to rediscovery of known compounds. Modern tandem separation/chemical characterization instrumentation can eliminate much of this problem by identi®cation of compounds before they are bioassayed.