Abstract: Residential air conditioning load serves as a crucial flexibility resource in the power system, and the evaluation of its regulation potential contributes to efficient load management and promotes energy conservation and emission reduction. However, due to the limits such as financial cost, data storage, and privacy protection, direct measurement of the residential air conditioning states and parameters is difficult, thereby hindering the differentiation evaluation of the air conditioning load regulation potential of different residents. Furthermore, residents use air conditioning to meet their thermal comfort requirements. Neglecting the constraints of thermal comfort when evaluating the air conditioning load regulation potential may result in overestimating evaluation values, thereby diminishing the effectiveness of load management strategies. Hence, this paper proposes a differentiation evaluation method of regulation potential for residential air conditioning load considering the thermal comfort. The method establishes the models for residential thermal comfort and air conditioning thermodynamics, and extracts key parameters that influence the residential thermal comfort and air conditioning load. Subsequently, a hidden Markov model and the Viterbi algorithm are employed to identify the air conditioning states and parameters. On this basis, the air conditioning load regulation potentials within the acceptable ranges of different resident thermal comfort levels are calculated. The case analysis based on real residential load data demonstrates that the proposed method accurately identifies the residential air conditioning states, and effectively evaluates the air conditioning load regulation potential that meets the thermal comfort requirements of residents.