Integration of sensing system and a split air conditioner (AC) gives better occupants' comfort control in a built environment. To reach optimal control from the air-conditioning system, various parameters such as temperature, humidity, and wind speed are used to indicate thermal comfort of the occupants. In this work, we proposed thermal-comfort control through an occupancy-detection algorithm and a fuzzy inference system (FIS). The proposed method aims to maintain occupant's thermal comfort between a comfort bound [0.5, −0.5] of Predicted Mean Vote (PMV) index. The occupancy-detection algorithm estimates a distance between the occupant and the split AC, which is an input variable for the FIS. The FIS use the estimated distance and the PMV index at the occupant's position to find the optimal set mode of the split AC in real time. Two conventional controls of the split AC, FIXED mode and AUTO mode, are compared with the proposed method. Our experimental results show that the FIXED mode and the AUTO mode can control the PMV index that are within the comfort bound for 13.19% and 80.22% of a total experimental time, while the proposed method achieves 100% of the total experimental time.