Determining soil ice content during freezing and thawing is important and challenging for both engineering and environmental issues. The thermo-time domain reflectometry (T-TDR) probe, which can monitor unfrozen soil water content and soil thermal properties simultaneously, has the potential to measure ice content in partially frozen soils. The objective of this study was to identify an optimum heat application strategy for measuring soil thermal properties with T-TDR probes in partially frozen soil while minimizing ice melting during the process. The optimized heating schemes were then applied for monitoring soil ice content dynamics during freezing and thawing. The results indicated that the heat pulse method failed at temperatures between −5 and 0°C because of temperature field disturbances from latent heat of fusion. When soil temperatures were ≤ −5°C, ice melting during heat pulse applications could be limited effectively with a combination of 60-s heat-pulse duration and 450 J m−1 heating strength, or a 90-s heat-pulse duration and heating strength of 450 to 900 J m−1. With the optimized heating scheme, T-TDR probes were able to measure soil ice content changes at ≤ −5°C during freezing and thawing, and the errors were within ±0.05 m3 m−3 in sandy loams and within ±0.1 m3 m−3 in soils with high clay content. At temperatures between −5 and 0°C, soil ice contents could not be measured accurately with the heat-pulse method directly, but they could be estimated coarsely from water content before freezing, TDR measured unfrozen water content, and T-TDR measured total water content at temperatures below −5°C.