Fluid percussion-induced traumatic brain injury models have been widely used in experimental research for years. In an experiment, the stability of impaction is inevitably affected by factors such as the appearance of liquid spikes. Management of impact pressure is a crucial factor that determines the stability of these models, and direction of impact control is another basic element. To improve experimental stability, we calculated a pressure curve by generating repeated impacts using a fluid percussion device at different pendulum angles. A stereotactic frame was used to control the direction of impact. We produced stable and reproducible models, including mild, moderate, and severe traumatic brain injury, using the MODEL01-B device at pendulum angles of 6°, 11° and 13°, with corresponding impact force values of 1.0 ± 0.11 atm(101.32 ± 11.16 k Pa), 2.6 ± 0.16 atm(263.44 ± 16.21 k Pa), and 3.6 ± 0.16 atm(364.77 ± 16.21 k Pa), respectively. Behavioral tests, hematoxylin-eosin staining, and magnetic resonance imaging revealed that models for different degrees of injury were consistent with the clinical properties of mild, moderate, and severe craniocerebral injuries. Using this method, we established fluid percussion models for different degrees of injury and stabilized pathological features based on precise power and direction control. Fluid percussion-induced traumatic brain injury models have been widely used in experimental research for years. In an experiment, the stability of impaction is inevitably affected by factors such as the appearance of liquid spikes. Management of impact pressure is a crucial factor that determines the stability of these models, and direction of impact control is another basic element. To improve experimental stability, we calculated a pressure curve by generating repeated impacts using a fluid percussion device at different pendulum angles. A stereotactic frame was used to control the direction of impact We produced stable and reproducible models, including mild, moderate, and severe traumatic brain injury, using the MODEL01-B device at pendulum angles of 6 °, 11 ° and 13 ° with corresponding impact force values ​​of 1.0 ± 0.11 atm (101.32 ± 11.16 kPa), 2.6 ± 0.16 atm (263.44 ± 16.21 kPa), and 3.6 ± 0.16 atm (364.77 ± 16.21 kPa), respectively. Behavioral tests, hematoxylin-eosin staining, and magnetic resonance imaging revealed that models for different degrees of injury were consistent with the clinical properties of mild, moderate, and severe craniocerebral injuries. Using this method, we established fluid percussion models for different degrees of injury and stabilized pathological features based on precise power and direction control.