Hydraulic engineering construction safety has become a major concern in engineering sustainability. Fall accidents, as a common type of accident during the hydraulic engineering construction process, have caused physical and fatal injuries and property losses on an individual and societal scale. With a sizable workforce, complex operational structures and demanding construction conditions, hydraulic engineering projects present more pronounced safety management challenges than other infrastructure initiatives. As a result, the risk of accidents, particularly fall accidents, is heightened in this domain. To prevent fall accidents and minimize losses, this study used the investigation reports of 389 cases of fall accidents as the analyzed corpus, and 16 contributing factors of fall accidents were extracted with the utilization of text mining. Accident feature terms were visualized through word clouds and ring bar graphs. The logical relationship among the influencing factors was quantified based on Decision-Making Trial and Evaluation Laboratory and Interpretative Structural Modeling (DEMATEL-ISM). The contributing factors and occurrence mechanism of fall accidents in hydraulic engineering were analyzed by establishing a multilevel hierarchical hybrid model. The results showed that the multilevel hierarchical hybrid model was divided into five levels. Thirteen causal chains were obtained. Chaotic security management, weak safety awareness and an inadequate safety system were the most critical factors, while the remaining eleven transitional factors and four surface factors also contributed significantly to the occurrence of accidents. Human and management factors dominated the overall factor transfer pathway. This study proposes countermeasures to the above-mentioned factors and provides a theoretical basis for the sustainable and safe construction of hydraulic engineering.