The adoption of soil fertility management practices (SFMPs) has become an important issue in the development economies, especially as a way to tackle land degradation, erosion, and low agricultural productivity. This study analyses the factors that facilitate or impede the probability and extent of adoption of multiple SFMPs as well as the performance effects, using farm survey data of 773 vegetable producers in rural China. Multivariate and ordered probit models are applied to the modeling of adoption decisions by farm households facing multiple SFMPs, which can be adopted in various combinations. A multinominal endogenous switching regression model is used to investigate the impact of SFMP adoption on farm productivity. The results show that: (1) the adoption of straw returning and advanced irrigation have substitution effect, and subsoiling practice is significantly correlated to straw returning and soil testing; (2) both the probability and the extent of adoption of SFMPs are influenced by many factors: household’s education, cadre membership, cooperative and training participation, social capital and individual awareness; (3) farms’ productivity is increasing with the intensive adoption of SFMPs. These results imply that policymakers should seek to promote local institutions and training providers, increase household education and awareness, and strengthen social networks in order to improve the adoption of SFMPs.