Background. The drowning death rate of Chinese students ranks among the top three in the world, and drowning has become an urgent problem for the country and society to solve. Objective. To test whether traditional swimming teaching can improve students’ knowledge and skills of water safety and reduce the probability of drowning in China. Methods. A total of 82 second‐grade students in elementary school were selected as experimental subjects, and the repeated measures analysis of variance experiment design was used to carry out the research. Results. After the experiment, the swimming safety education model significantly increased the scores of swimming safety knowledge and skills and the difference was statistically significant (t = 4.060, 5.325, P < 0.05); the attitude and behavior scores decreased, and the difference was statistically significant (t = −4.392, −2.201, P < 0.05). After the experiment, the swimming safety education model is more effective in teaching swimming safety knowledge, attitude, and behavior than the traditional swimming teaching, and the difference is statistically significant (t = 2.699, 3.852, 4.308, P < 0.05). After the experiment, when the traditional swimming teaching model is compared with the control group, the difference in swimming safety skills was statistically significant (t = 4.704, P < 0.05) and other dimensions were homogeneous. In conclusion, the traditional swimming teaching can only improve student’s swimming safety skills but cannot reduce student’s drowning injuries. The swimming safety education model is superior to the traditional swimming teaching in terms of swimming safety knowledge, attitudes, and behaviors. Recommendations. For beginners in swimming, the teaching content of swimming self‐rescue skills and swimming safety knowledge should be added.