Current biological control strategies involve the use of exotic or nat
ive biocontrol agents, proven to be host-specific and lethal. This is
a successful field of endeavor, attractive to a large cadre of entomol
ogists and a few plant pathologists. The main limitation is that each
weed species must be attacked by a separate host-specific agent and th
ese organisms are not always lethal. Only by using a radical paradigm
shift can we reach a new strategy of biocontrol. In this strategy, let
hal, broad host-range pathogens are genetically modified to permit the
ir safe release. Either they are rendered host-specific or they are gi
ven a chemical dependency that prevents their spread or long-term surv
ival. This genetic-manipulative approach offers numerous and diverse s
cenarios for biocontrol. Host-specific promoters or toxins, host-depen
dency by multiple auxotrophy, or mutants dependent on specific environ
mental conditions are all approaches that lend themselves to genetic m
odification. Such self-delimiting pathogens may have the advantage of
repeated marketability, environmental safety and multiple target weeds
.