Many national parks in Australian rangelands were formerly grazing properties. Even after domestic stock has been removed, the landscapes, especially those near artificial watering points, remained degraded with little or no regeneration of native plants. Congregating large macropods have often been blamed for impeding regeneration of plant communities and rehabilitation of these degraded landscapes. Following a severe drought which reduced densities of large macropods very significantly in Idalia National Park, central-western Queensland, I investigated whether closure of watering points would hamper their recolonisation and influence their grazing distribution and forage selection in the park. I also determined the spatial and temporal pattern of degradation around the watering points, using a suite of indicators. A BACI (Before After Control Impact) type experiment was conducted using two replicate watering points each for impact (fenced) and control (unfenced) treatments. Two watering points were fenced off behind a kangaroo-proof fence in August and December 2002, respectively, in the deepening drought. Walked 4 km line transects were used to estimate densities of red kangaroos (Macropus rufus) and wallaroos (M. robustus) in areas within a 4 km radius of each watering point. The accumulation of dung pellets was used to estimate the grazing distribution of large macropods around water. Abundance and composition of vegetation were monitored on fixed transects using a 0.5 * 0.5 m quadrat 0.5 to 4 km from each watering point. Herbivore offtake was monitored using a movable cage technique. Measures of landscape function (stability, infiltration and nutrient rating) and ant composition were monitored at various distances from each watering point. Monitoring was commenced two to six months before the closure of watering points and completed 21 and 25 months following the closure, every two to five monthly to follow changes in large macropod densities and vegetation communities. Prior to the drought, densities up to 130 large macropods km-2 were recorded in the park. The average density of wallaroos declined by 97.7% during the drought whereas that of red kangaroos declined by 83.8%. The first substantial rain fell in January 2003, but rainfall throughout the study period continued to be much lower than the long-term median, except for January 2004 when more than half the average annual rainfall fell, causing major flooding. Closure of watering points coincident with the drought inhibited red kangaroo recolonisation by 40% in the areas surrounding fenced watering points for up to 21 months following the closure, with distances to alternative watering points being 3.8-6.6 km. Red kangaroos in an area surrounding one of the unfenced watering point recovered to the pre-drought average density. The grazing distribution of large macropods within 4 km from the watering points, however, was determined primarily by where plentiful food resources were, not by the location of water. Large macropods in the areas without access to water did not select forage with higher water contents, compared to those in the areas with water access. Differences in their forage preference between the two areas appeared to reflect the difference in vegetation composition between the areas. Sida filiformis, Cynodon dactylon and Tragus australianus, were identified as grazing sensitive plant species. In some cases, closing watering points helped these plants to increase their abundance following the drought. Most indicators (the percentage of bare soil, biomasses of perennial grasses and round-leaved chenopods, ant functional groups and the measures of landscape function) showed that the exclusion of mammalian grazing for more than a decade from long-term exclosures had facilitated regeneration of perennial grasses and landscape rehabilitation at least at some sites. At one long-term exclosure, where topsoil and seed bank had presumably been lost already, no regeneration occurred. When several indicators were used to determine land conditions around watering points, their responses were not always consistent. Nevertheless, the overall spatial patterns of degradation did not match the spatial pattern of grazing distribution by large macropods around water. Therefore, the current degradation appears to be a legacy of the livestock overgrazing that had continued for more than 100 years prior to the park establishment. The large drought-induced reduction in the large macropod densities did not result in any conspicuous reversal of the patterns of degradation around the watering points.