The objective of this paper is to test three main hypotheses in grassland succession by using 17-year observational data: (i) species diversity increases during early pioneering stages; (ii) the similarity rate of experimental plant communities increases along with the succession process (dissimilarity rate among plots decrease with succession age); and (iii) plant communities in different field plots converge towards a quite similar composition during the initial years of succession. Results draw from quantitative statistics which including: the general development of plant communities, Shannon's H and E, Bray-Curtis dissimilarity and community turnover, indicate that after first 5 years of succession, species diversity rapidly becomes alike among all the plots. The average species number in all plots increased constantly from 8 to 25. Meanwhile, the total species number for all plots increased constantly from 23 to 55, accompanied by a steady decrease in Standard Deviation (S.D.) from 4.6 to 0.3. Additionally, dissimilarity of all stands in species composition decreased, indicating a clear trend towards a rapid convergence in species richness and composition. The most important effect of cessation of weeding is that the artificial diversity gradient from 1 to 16 species gets diminished by immigration in species poor plots.