In order to fastly, accurately and non-destructively to predict the biomass on alpine typical steppe in Qilian Mountains, growth indexes of plants species such as plant height, coverage were observed during the growing season under two different pastures, which are Winter grazing land and spring and autumn grazing land. In total 60 quadrats were measured. Taking the Plant height, coverage, the product of plant height and coverage of 6 major species in winter pasture,such as Stipa purpurea, Achnatherum inebrians, Leymus secalinus, Agropyron cristatum, Potentilla bifurca and Convolvulus ammannii as independent variable, and the main population, other populations and aboveground biomass of the community in the same quadrat were used to develop the predicted model, which was verified and corrected by applying the collected data under Vegetation data of Spring and Autumn Pasture. The results showed that the growth indices of representative species can predict the aboveground biomass of themselves, other species and communities. The composite factor of plant height and coverage product was the best for predicting the population. The single factor of plant height and coverage was better than the compound factor in predicting the other populations of the four grasses and the compositae populations of the Potentilla bifurca. Six main species alone or two to six kinds of combining both predictable community biomass on the ground, but the product of 6 species of plant height and coverage to participate in the prediction of regression curve at the same time, decision coefficient is the largest, can explain community biomass of 89.50% of the variation on the ground, can be used as a typical alpine steppe community biomass optimal prediction model on the ground.