High manganese steels can damage the differential thermal analysis (DTA) instrument due to the manganese evaporation during high temperature experiments. After analyzing the relationship between residual oxygen and manganese evaporation, tantalum metal was employed to modify the crucible of DTA, and zirconium getter together with strict gas purification measures were applied to control the volatilization of manganese. By these modifications, problems of thermocouple damage and DTA instrument contamination were successfully resolved. Cobalt samples were adopted to calibrate the accuracy of DTA instruments under the same trial condition of high manganese steel samples, and the detection error was confirmed to be less than 1 °C. Liquidus and solidus temperatures of high Mn steels were measured by improved DTA method. It was found that the liquidus temperatures of samples tested by experiments increased linearly with the heating rates. To eliminate the effects of the heating rate, equilibrium liquidus temperature was determined by fitting the liquidus temperatures at different heating rates, and referred as real liquidus temperature. No clear relationship between solidus temperatures and heating rates was found, and the solidus temperature was finally set as the average value of several experimental data.