Induced Polarization method (IP) is the leading method of mining geophysics because it allows detecting disseminated ores, which frequently can serve as indicator of economically valuable deposits. In Russia, most of mining districts actively investigated with IP method, are located in Far East and Eastern Siberia. At such cold regions rocks are temporarily or permanently frozen. However, to date, only scares data concerning IP of frozen disseminated ores are available. In this paper we studied low-frequency (from 10 mHz to 100 Hz) spectral IP (SIP) of synthetic models of disseminated ores made with a mixture of sand and cryptomelane. We carried out SIP measurements in the temperature range from +10^0С to -25^0С in the course of successive decrease and increase of the temperature. In addition we measured SIP response at the reference model made with the sand alone. Four regions were detected in the relationship between the absolute value of the complex resistivity and temperature. With decrease of the temperature (1) the resistivity slightly increases at the positive temperature range following the linear trend; (2) the resistivity strongly increases when the temperature goes down from zero to -2 – -3 ^0С; (3) with further increase of the temperature up to -10 – -15 ^0С the resistivity continues to increase with slower rate, following an exponential trend; and (4) with further increase of the temperature up to 20 0 C the resistivity follow again the exponential trend, but with smaller coefficient of the exponent. In the third region we detected a hysteretic behavior of the ‘resistivity – temperature’ relationship. We obtained very similar results with the samples contained the water saturated sand alone. With increase of the temperature the maximum value of the complex resistivity phase decreases, and, in contrast, the IP relaxation time increases. These relationships also show hysteresis. We explain these hysteretic behaviors of the studied parameters by (1) a decrease of the water saturation in the course of water freezing, (2) a rise of salt into pore water from appeared ice phase, and (3) a different dynamics of ice melting and water freezing. We believe the aforementioned phenomena lead to different tortuosity of the liquid phase when the temperature decreases and increases. Our data show that the IP magnitude characterized by the complex resistivity phase, decrease twice comparing to the unfrozen sample, when the sample temperature is about -10^0С. We believe that at - 10^0С the free water is completely frozen. We state, therefore that at negative temperature the IP effect does not disappear, but their magnitude strongly reduces, which must be considered when planning IP surveys and analyzing IP data at cold regions.