Water Quality Improvement Plans (WQIPs) are being developed for individual river basins on the Great Barrier Reef (GBR) catchment associated with the GBR Water Quality Protection Plan. Within each WQIP, marine ecosystem targets are linked to end-of-river pollutant (suspended sediments, nutrients and pesticides) load targets and to farm level management practice targets. The targets are linked through quantitative models; e.g. one model connects GBR chlorophyll concentrations (marine target) to end-of-river nitrate loads, a second connects the end-of-river nitrate loads to fertiliser management targets in the catchment, whereas a third model links fertiliser application to nitrate loss at the farm scale. The difficulties of applying these linked models to derive credible and practical management targets are great, given the high degree of uncertainty in each model. Our understanding of the generation of suspended sediments, nutrients and pesticides in catchments and the relationship to on-farm management, the transport of these materials to the ocean, their transport in coastal waters and their effects on marine ecosystems is incomplete. The challenge is to produce estimates from the models, with known levels of uncertainty, but robust enough for management purposes. Case studies from the Tully–Murray basin and the Burdekin basin in north Queensland are discussed.