Traditional methods for evaluating exercise intensity have problems such as insufficient accuracy, difficulty adapting to individual differences, and dynamic changes. This paper studies an exercise intensity evaluation technique based on fuzzy observer. The fuzzy observer adjusts the rule weights in response to dynamic changes with the help of the real-time adjustment mechanism, bringing the assessment algorithm closer to the individual’s physiological state and the state of the environment at the moment. In the experiment, the estimation performance under the receiver operating characteristic (ROC) curve was noteworthy, with accuracy and precision reaching 0.93 and 0.91, respectively, at a threshold of 0.52. These results point to an adaptive approach to the measurement of exercise intensity.