Abstract: Stipa purpurea alpine grassland is the most representative and important natural pasture of Northern Tibet but in recent years has suffered large-scale degradation, mainly from overgrazing. Optimization of S. purpurea alpine grassland grazing systems in Northern Tibet is a very important issue for local policy makers. Livestock feed intake is an evaluation method of grassland productivity and is one of the important tools of grassland management as it is a key factor in the grazing system. Grassland standing crops and the feed intake rate of Tibetan sheep are critical criteria for healthy and sustainable development of alpine grassland in Northern Tibet. Based on 5 years (2006-2010) of grazing in S. purpurea alpine grassland in Amdo County, Nagqu Region, Tibet Autonomous Region, the biomass of standing crops and net primary production of grassland and Tibetan sheep feed intake were studied using four grazing rates. In the first three years (2006-2008) of the experiment, the stocking rate was: 0 sheep/1.67 ha (CK), 2 sheep/1.67 ha (LG, light grazing), 3 sheep/1.67 ha (MG, moderate grazing), and 5 sheep/1.67 ha (HG, heavy grazing). In a later experiment (2009-2010), the grazing rate was doubled to 0 sheep/1.67 ha (CK), 4 sheep/1.67 ha (LG), 6 sheep/1.67 ha (MG), and 10 sheep/1.67 ha (HG). In the growing season of 6-9 months, based on the method of community investigation, the standing crops within 0.5 m×0.5 m quadrants were measured and the feed intake was calculated by the formula: Feed intake=Before grazing biomass-after grazing biomass+growth matter. There was no significant effect on standing crops of grazing intensity in the first study, but with the doubling of grazing intensity in the later experiment, the standing crops decreased with increased grazing intensity. The aboveground net primary production of the grazing area was larger than in the control pilot in 2006 but by 2010, in the control treatment it was significantly (P<0.05) larger than in the grazing treatment. The feed intake of Tibetan sheep was significantly negatively correlated with grazing intensity (P<0.05). The order of intake rate of sheep was: heavy grazing intensity>light grazing intensity>moderate grazing intensity. Comprehensive analysis of the results indicated that, in the light grazing treatment, not only was the biomass of standing crops and aboveground net primary production the largest, but also the feed intake of sheep was relatively high. We therefore consider that the optimal stocking rate of S. purpurea alpine grassland in Northern Tibet should be less than 4 sheep/1.67 ha.