The two approaches, classical and inundative, which can be used to exploit fungi for weed biocontrol, are described and discussed. It is concluded that classical biological control involving the release of coevolved natural enemies, typically highly specific biotrophs into an exotic ecosystem, is the most environmentally and economically acceptable long-term strategy for controlling alien weeds, particularly in conservation areas. Mycoherbicides, based mainly on endemic nectrotrophs, have still not realized their full potential, mainly owing to problems with formulation, application, and market size, but with improved technology these may play an increasingly significant role in agricultural ecosystems. Virulent pathogens with broad host ranges offer the most commercial potential, if these can be tailored to suit individual crop systems. The initial selection of suitable biocontrol agents, based on field, herbarium, and literature surveys is outlined, whilst the problems involved in making the final selection are analysed in detail, especially those relating to taxonomy, biology, and host specificity. Examples are given of fungal pathogens that are being or could be employed for control of some major tropical weed species. Key words: fungi, classical biocontrol, mycoherbicides, tropics, conservation.