To compare the effects of different sampling techniques for alpine grassland and improve the efficiency and accuracy of alpine grassland plant diversity study, we applied four different methods to investigate the diversity in three main types of grassland in QinghaiTibet Plateau (alpine meadow, alpine steppe and alpine desert). The methods included centerstaked transectquadrat method, Daubenmire transect method as modified by the USDA Forest Service, large quadrattransect method proposed by the USDA Agricultural Research Service, and modified Whittaker multiscale vegetation plot method. Through analysis of monitoring effectiveness and cost efficiency of the three kinds of alpine grassland plant diversity, we screened out the investigation method which had the highest accuracy and lowest cost.  Species numbertime cost curves indicated that the difference of consumption time of four methods became larger as the species number in sampling plots increased.  In alpine meadow and alpine steppe, which had relatively higher species richness, the modified Whittaker multiscale vegetation plot method can detect the highest species richness; however, this method costs the longest time. Relatively, the centerstaked transectquadrat method was the most economical.  In alpine desert with low species richness, the time cost of investigation and the accuracy of species diversity did not vary much among the four investigation methods. As a whole, considering the simplicity, large quadrattransect method was the best one for investigation.