A field experiment was conducted to study biomass allocation and compensatory growth of phalanx clonal plants under different grazing disturbance pattern and nutrient availability. We also tested foraging models for clonal plants and some hypotheses related to compensatory growth patterns, such as the grazing optimization hypothesis (GOH), the continuum of responses hypothesis(CRH), and the growth rate model(GRM). We examined an alpine meadow clonal plant Kobresia humilis in a field grazing defoliation experiment by domestic animals through setting grazed/ungrazed pots. There were 3 habitats, (Ⅰ) pen pasture, with heavy defoliation ratio×high nutrient availability. (Ⅱ) pass pasture, with moderate defoliation ratio×low nutrient availability. (Ⅲ) ungrazed pasture, with null defoliation ratio×low nutrient availability. Our results showed that regardless of release from grazing defoliation in the current year, the more the risk of defoliation encountered by plants in spring, the lower the biomass allocation to growth organ (leaves), and the greater the biomass allocation to storage organ (rhizomes). The reproductive allocation was highest in the conditions of moderate defoliation ratio and low nutrient availability. However, biomass allocation to different parts of the ramet did not differ between grazed and ungrazed treatments. Overcompensation was found in habitat 2, whereas fullcompensation occurred in habitat 1. Moreover, in habitat 1 ramet density significantly decreased. The pattern of biomass allocation in the species was not consistent with the foraging model prediction. The pattern of compensatory growth supported the prediction of GOH and GRM, however, it did not agree with CRH. These results suggested grazing defoliation significantly affect the patterns of biomass allocation and compensatory growth of K. humilis ramet. Moderate defoliation causes the overcompensation. But heavy defoliation will have a negative effect on long-term persistence of the population in the area.