Abstract: Based on the data of longterm fertilization experiment in red paddy soil at Jinxian county, Jiangxi province. Integrated soil fertility was evaluated and its effects were analyzed by fuzzy mathematics and statistics methods. The results showed that organic matter, available N, available P, available K, physical clay, aggregate ratio and aggregate stability were proposed a minimum data set(MDS) for soil fertility assessment and the evaluation results produced by MDS had good representativeness and scientificalness. Among all treatments, SFI for NPKM was maximum and had a significant difference with SFI for other treatments. Compared to the control, the SFI of NPKM, 2NPK and NPK treatments was about 31.3%, 13.0% and 10.1% higher respectively. As time went on, the SFI for NPKM treatment increased firstly and decreased subsequently and the critical time was 12th year after cultivated, the SFI for N and CK treatments slowly decreased and that of other treatments slowly increased over time. Within a certain range, the quadratic curve fitting could better reveal the relationship between total yield and SFI than that of the linear fitting. Further fitting found that the relationship of SFI and relative yield accorded with “S” shape Boltzmann growth model, from this a conclusion could be drawn that when SFI increased to 0.60, the increment of relative yield was close to zero, and NPKM treatment (SFI=0.688) had demonstrated the obvious similar effect. When relative yield closed to stable, corresponding fertilization age was the years of 2 (NPKM), 8(2NPK), 9(NPK), 11(NP), 13(K and NK) and 14(P), respectively. In summary, longterm balanced fertilization (NPKM, 2NPK and NPK) had improved obviously integrated soil fertility, especially NPKM treatment, but its present situation had no effect on increasing crop yield and fertilizer application should be reduced or prohibited for several years. Longterm partial fertilization could slightly increase SFI, but longterm single N application made SFI decrease.