To measure the thermal conductivity of liquids using a hot probe, it is necessary to minimize the disturbance caused by natural convection. With this in mind, two options are investigated in this work. One is to shorten the measurement duration; the other is to keep the power input (or the temperature rise of the probe) as low as possible. The measurement duration is studied using the three-ply composite heat transfer model, which includes the heater, the guard body and the measured sample. A resistance bridge is introduced to improve the measurement precision and the maximum temperature rise of the probe can be kept within 1 °C. According to the experimental results on liquid samples, the measurement time interval (MTI) is about 5–20 s, and the measurement uncertainty is about 3%. The thermal conductivities of dimethyl sulfoxide (DMSO) and its aqueous solutions have been determined using the proposed technique, and may not have been reported previously in the literature.