The influence of the chemical composition of a NiTi alloy on the martensite stabilization effect was studied. The Ni- 50.0 at. %Ti, Ni – 49.5 at. % Ti and Ni – 49.0 at. %Ti alloys were quenched from 900 °C (10 min) into water and after this heat treatment, the alloys underwent the B2 ↔ B19′ transformation on cooling and heating without the R phase formation. The martensite stabilization effect was observed in NiTi alloys regardless of the chemical composition and value of the preliminary strain. The value of the martensite stabilization effect was measured as the difference between the temperatures that were measured during the first and the second heating. When the residual strain was less than 2.5%, the martensite stabilization effect values were close to each other in all studied alloys. Otherwise, if the residual strain exceeded 2.5%, the martensite stabilization effect values in the Ni- 50.0 at. %Ti and Ni – 49.5 at. % Ti alloys were larger than in the Ni – 49.0 at. %Ti alloy. It was shown that there was no correspondence between the values of the martensite stabilization effect and the irreversible plastic strain that appeared in the samples during the preliminary deformation. The martensite stabilization effect was found in the Ni – 49.0 at. %Ti alloy after preliminary deformation up to 10% or less that was not accompanied by a plastic strain. Thus, it was shown that the plastic strain was not the main reason for the martensite stabilization effect. A new hypothesis was assumed that a loss in the coherency of the interface that was caused by the martensite reorientation and detwinning during the preliminary deformation might be responsible for an increase in the temperatures of the reverse transformation that occurred on the first heating.