Soil ice content (theta(i)) is an important property for many studies associated with cold regions. In situ quantification of theta(i) with thermo-time domain reflectometry (TDR) at temperatures near the freezing point has been difficult. The objective of this study is to propose and test a new thermo-TDR approach to determine theta(i). First, the liquid water content (theta(l)) of a partially frozen soil is determined from a TDR waveform. Next, a pulse of heat is applied through the thermo-TDR sensor to melt the ice in the partially frozen soil. Then, a second TDR waveform is obtained after melting to determine the theta(l), which is equivalent to the total water content (theta(t)) of the partially frozen soil. Finally, theta(i) is calculated as the difference between theta(t) and theta(l). The performance of the new approach was evaluated in sand and loam soils at a variety of theta(t) values. The new approach estimated theta(t), theta(l), and theta(i) accurately. The root mean square errors (RMSE) of estimation were 0.013, 0.020, and 0.023 m(3) m(-3) for sand, and 0.041, 0.026, and 0.031 m(3) m(-3) for loam. These RMSE values are smaller than those reported in earlier thermo-TDR studies. Repeating the thermo-TDR measurements at the same location on the same soil sample caused decreased accuracy of estimated values, because of radial water transfer away from the heater tube of the thermo-TDR sensor. Further research is needed to determine if it is possible to obtain accurate repeated measurements. The use of a dielectric mixing model to convert the soil apparent dielectric constant to theta(l) improved the accuracy of this approach. In our investigation, application of a small heat intensity until the partially frozen soil temperature became larger than about 1 degrees C was favorable. The new method was shown to be suitable for estimating ice contents in soil at temperatures between 0 degrees C and -2 degrees C, and it could be combined with the volumetric heat capacity or thermal conductivity thermo-TDR based methods, which measured ice content at colder temperatures. Thus, the thermo-TDR technique could measure theta(i) at all temperatures.