Current methods that utilize simple data or models to judge whether soil fertility can selfdevelop are not sufficiently rigorous. A new framework has been set up using catastrophe theory, laboratory experiment, field work, and 3S(Geographic information system, Global positioning system, and Remote sensing) to explore soil fertility catastrophe under ecological restoration, discriminate whether soil fertility can self-develop, and propose adjustment of ecological restoration measures in the Zhuxi watershed of Changting County, Fujian Province, China, which is a typical representative of the red soil hilly region of China. The results show that: 1) the soil fertility is obviously improved through the four ecological restoration measures, which impels soil fertility catastrophe. Among 89 soil samples, catastrophic soil samples and stable soil samples account for 26(29.21%) and 63(70.79%) of the samples, respectively. The four ecological restoration measures are listed in the order lowquality forest improvement > arbor–bush–herb mixed plantation > orchard improvement > closing measures according to the proportions of catastrophic soil samples. A typical soil sample in Bashilihe that can self-develop is selected as the criterion to judge the upper lobe and lower lobe of soil fertility in the process surface of the Cusp catastrophe model. Twenty-six(29.21%) were in the middle lobe, 10(11.24%) were in the upper lobe, and 53(70.79%) were in the lower lobe. The catastrophic direction of 26 catastrophic soil samples is to the upper lobe according to soil and water loss change as well as fieldwork. There is a significant positive correlation of Δ with soil and water loss change, and the lower soil and water loss relates to higher catastrophic probability. 2) Soil fertility self-development could be regionalized as “Soil fertility can self-develop” whose area was 12.74 km2(28.33%) distributed mainly in the leftmost and rightmost parts, “Soil fertility tends to self-develop” whose area was 11.63 km2(25.89%) distributed mainly in the middle part, and “Soil fertility cannot self-develop” whose area was 20.58 km2(45.78%) distributed mainly between the above two types. 3) There is no need to take ecological restoration measures and excessive human interference should be avoided in the future in regions of “Soil fertility can self-develop” and “Soil fertility tends to self-develop,” and ecological restoration measures should be taken in region of “Soil fertility cannot self-develop.” 4) We suggest withdrawal and implementation of ecological restoration measures should be incorporated into the evaluation criteria of ecological restoration to avoid misuse of funds. Current methods that utilize simple data or models to judge whether soil fertility can selfdevelop are not verified rigorous. A new framework has been set up using catastrophe theory, laboratory experiments, field work, and 3S (Geographic information system, Global positioning system, and Remote sensing to to soil fertility catastrophe under self-develop, and propose adjustment of soil restoration measures in the Zhuxi watershed of Changting County, Fujian Province, China, which is a typical representative of the red soil hilly The results show that: 1) the soil fertility is obviously improved through the four ecological restoration measures, which impels soil fertility catastrophe. Among the 89 soil samples, catastrophic soil samples and stable soil samples account for 26 (29.21%) and 63 (70.79%) of the samples, respectively. The four ecological restoration measures are listed in the order lowqual ity forest improvement> arbor-bush-herb mixed plantation> orchard improvement> closing measures according to the proportions of cat proportions of catastrophic soil samples. A typical soil sample in Bashilihe that can self-develop is selected as the criterion to judge the upper lobe and lower lobe of soil fertility in the process surface of the Cusp catastrophe model. Twenty-six (29.21%) were in the middle lobe, 10 (11.24%) were in the upper lobe, and 53 catastrophic direction of 26 catastrophic soil samples is to the upper lobe according to soil and water loss change as well as fieldwork. There is a significant positive correlation of Δ with soil and water loss change, and the lower soil and water loss due to higher catastrophic 2) Soil fertility self-development could be regionalized as “Soil fertility can self-develop ” whose area was 12.74 km2 (28.33%) distributed mainly in the leftmost and rightmost parts, “Soil fertility tends to self-deve lop ”whose area was 11.63 km2 (25.89%) distributed mainly in the middle part, and “Soil fertility can not self-develop ” whose area was 20.58 km2 (45.78%) distributed mainly between the above two types. to take ecological restoration measures and excessive human interference should be avoided in the future in regions of “Soil fertility can self-develop ” and “Soil fertility tends to self-develop, ” and ecological restoration measures should be taken in region of soil fertility can not self-develop. "4) We suggest withdrawal and implementation of ecological restoration measures should be incorporated into the evaluation criteria of ecological restoration to avoid misuse of funds.