Lantana camara (lantana), a thicket-forming shrub, a number of different varieties of which were introduced into South Africa as ornamental plants but which has become a serious invasive weed. Conventional control measures for lantana are expensive and ineffective and it has therefore been targeted for biological control since 1961. To date, eleven biological control agent species have become established on lantana in South Africa. However, most agents persist at low densities and only occasionally impact plant populations. Three species regularly cause significant damage, but only reach sufficiently high numbers by midsummer after populations crash during the winter. Overall, the impact of the biological control programme on the weed is negligible and this has been ascribed to the poor selection of agents for release, the accumulation of native parasitoids, differences in insect preference for different varieties of the weed and variable climatic conditions over the weed’s range. This study suggests that the importance of varietal preferences has been over-estimated. A predictive bioclimatic modelling technique showed that most of the agents established in South Africa have a wide climatic tolerance and that the redistribution and importation of new climatypes of these agents will not improve the level of control. Additional agents are required to improve the biocontrol in the temperate conditions, and also to increase damage in the sub-tropical areas where most of the agents are established and where the weed retains its leaves year round. New candidate agents that possess biological attributes that favour a high intrinsic rate of increase, a high impact per individual and that improve the synchrony between the weed and the agent in climatic conditions that promote the seasonal leaflessness of plants should receive prior consideration. A survey in Jamaica indicated that additional biological control agents are available in the region of origin but that care should be taken to prioritise the most effective agents. The various selection systems currently available in weed biocontrol produce contradictory results in the priority assigned to candidate agents and a new selection system is proposed. The biology and host range of two new candidate natural enemies, the leaf-galling weevil, Coelocephalapion camarae and the leaf-sucking mirid, Falconia intermedia were investigated for the biocontrol of lantana. The studies indicated that these have considerable biocontrol potential, in that the weevil has a wide climatic tolerance and has the potential to survive the host leaflessness typical of temperate conditions, while the mirid has a high intrinsic rate of increase, and the potential for several generations a year. Both agents caused a high level of damage to the leaves, with the weevil galling the vascular tissue in the leaf-petiole and the mirid causing chlorotic speckling of the leaves. During laboratory trials both agents accepted indigenous species in the genus Lippia. However, under multiple choice conditions these agents showed a significant and strong oviposition preference for lantana. A risk assessment and post release field trials indicated that F. intermedia is likely to attack some Lippia species in the presence of lantana, but the levels of damage are predicted to be relatively low. A possible low incidence of damage to indigenous species was considered a justifiable ‘trade-off’ for the potentially marked impact on L. camara. Preference and performance studies on the two candidate agents suggested that most of the South African lantana varieties are suitable host plants. The mirid preferred certain varieties in multiple choice experiments, but this is unlikely to affect its impact under field conditions. Permission for release was accordingly sought for both species. Finally, the challenges facing the biological control programme and the potential for improving the control of L. camara in South Africa are considered.