Based on the research on the exercise capacity influencing factors of 100m sprint, this paper proposes comprehensive evaluation and prediction model based on fuzzy neural network, which is the improvement of the existing athletic ability evaluation systems and overcomes the shortcomings of traditional methods’ strong subjectivity. The obtained fuzzy neural network system cannot only ensure continuous learning and correction, but also can make the network converge quickly according to the practical significance of the specific parameters. Through examples test, the evaluation results are objective and have the ideal precision. The model can not only be used to simulate and predict the athletic performance, as a basis to develop targeted training programs, but also can provide quantitative criteria in the athletes selection process, which has a high application value and broad application prospects