Introduction: Droughts and flooding occur frequently due to climate change and human activities, which have significantly affected the ecological environment of farmland and crop production. Rock wool (RW) has some properties like high porosity and water retention capacity, and it is widely used in green roofing and agricultural production to reduce flood and drought disaster. Methods: We set 24 artificial rainfall experiments to analyze the impact of RW on the farmland runoff, soil water storage capacity (SWSC), nitrogen and phosphorus loss and crop growth. Finally, the Critic-Entropy comprehensive evaluation method was used to select the best solution for RW embedding. Results: The result shows that RW could reduce the runoff by 49.6%–93.3%, and it made the SWSC increase by.2%–11% Vol in the 10–70 cm depth. During the runoff process, the concentration of nitrogen and phosphorus decreased with the increase of the RW volume, while the nitrogen and phosphorus loss reduced by 51.9%–96.6% and 72.4%–96.4% respectively when RW was buried in the farmland, so RW could effectively promote soil and water conservation. Finally, RW increased the plant biomass and yield by a maximum of 12.1% and 20.4% respectively due to the large retention of water and nutrients. Therefore, combined with the above experimental result, this study indicates that RW could obtain the best comprehensive benefit in the embedding volume of 536.73 m3/ha. Discussion: When the volume of RW embedding was too large, RW easily caused insufficient water storage in short-duration rainfall or irrigation, which made the RW unable to release water to alleviate the soil water deficit during the critical growth period, which may result in drought events being aggravated and crop yield reduced. Overall, this study is conducive to clarifying the comprehensive application effect of RW in agricultural production, pollution control, and urban landscape, and it provides an important basis for expanding its application field and promoting the stability of farmland ecosystems. Copyright © 2023 Li, Liu, Wang, Qin, Zhang, Li and Feng.