In a field experiment we evaluated the effects of exclosure and sand-fixing measures on land restoration of shifting sand dunes in a semi-arid, sandy grassland located in Northern China. We fenced an area of about 2.1 ha, enclosing 2 shifting sand dunes-one large and the other small. In 1996, on each dune we applied 2 sand-fixing measures developed in China-burying wheat straw in a checkered pattern (Sc) and planting seedlings of a sand-fixing shrub, Artemisia halodendron (Ar). Changes in the topographical features and soil properties were then monitored for 6 years. The effectiveness of the sand-fixing measures was assessed by monitoring the topographical features and soil properties. Sand dune fixation took 1 year, and a biological soil crust with high contents of soil organic carbon and fine particles (clay+silt) was formed within 3 years. These results are consistent with those reported in previous studies, confirming that such crusts reduce wind erodibility, The effectiveness of exclosure alone, without the other measures, was also implied, especially for the small dune, based on the changes in the topographical features and soil properties outside the exclosure, where a "no-grazing" policy had been implemented from 2000. The rate of soil restoration was higher on the small dune than on the large dune. This was due to the differences in the initial soil conditions caused by the topographical features, suggesting that the capacity of a site to reverse desertification by itself depends on the dune size. Sc was slightly more effective than Ar, presumably because Ar inhibited the invasion of other plant species, whereas Sc allowed invasion of various species. Although the soil properties in the very thin surface crust were restored, the restoration of the soil properties in deeper layers was not significant even after 6 years. It would take a longer time (e.g., 10 to 20 years) for the restoration of the soil properties in deeper layers to sustain more intensive land use in this region.