Heat stress seriously affects the productivity, fertility and welfare of dairy cows. It is to be expected that, in the course of the next few decades, climate conditions for raising cattle will deteriorate. In order to reduce the risks of heat stress in dairy cows, researchers have been trying to seek a reliable method to predict or evaluate heat stress, and it has been become to a hot topic in this research field to obtain an index model by combining multiple environmental factors. To avoid blindness in selecting these environmental factors, this paper systematically analyzed the temperature-humidity index, and its modified indices based on THI combined with other factors, and compared the differences among these indices in characterizing the heat stress of dairy cows. These differences relate to the thresholds of indices, environmental factors involved and heat exchanging properties, and the relationships between physiological responses of dairy cows and indices. In this paper, the problems of heat stress index of dairy cows are also discussed. The most widely used temperature-humidity index is based on thermal sensation and thermal comfort of human, but had not been actually related to cow's living environments. Most indices linking environmental parameters to the physiological or behavioral responses of cows were established through the statistical correlation analysis process, but it did not reflect the real physical significance of the effects of environmental parameters on the process of heat exchange between animals and its environmental factors. In addition, the indices could only be well applied to the situation where they were established properly with the sufficient information for their use. We believed that THI had played an important role in the past when environmental parameters, physiology indicators and behavior responses were hard to be obtained, but the problem of THI had been existing in its application, so other direct parameters such as heat tolerance of high-yielding cows, the change of building structure and the linkage of various cooling modes should be considered when a new THI is being designed. Otherwise, the current THI will be not to meet the future production requirements of the dairy farming industry. With the application of digital technology, multi-parameter acquisition technology has been fundamentally changed, so it is necessary to develop a more reliable index of evaluating heat stress for cows to meet the needs of increasing livestock productivity and achieving welfare. Our suggestions to achieve this goal are as follows: 1) The index should be developed to incorporate more environmental parameters reflecting heat transfer mechanism. 2) The index can be applied not only in some specific climatic conditions, but also in other climatic conditions. 3) The index is to have applicable conditions and thresholds, and the thresholds of the index should be able to be dynamically adjusted to extend the scope of its application. 4) The index should be linked to animal physiological parameters and environmental factors, and consideration could be given to constructing a new index by meeting the theory of animal thermal balance.