This paper briefly introduces twenty one versions of garnet biotite Fe Mg exchange geothermometry, points out the sources of systematic errors inherent in geothermometry. The sources lie in that: (1) ideal Fe Mg mixing in garnet and biotite is assumed; (2) non ideal properties of both garnet and biotite are not considered completely; (3) minor elements—Mn, Ca in garnet, and Al Ⅵ, Ti, Mn in biotite are almost neglected; (4) effects of pressure on equilibrium, although little, are almost not taken into consideration, and experiments were conducted at one fixed pressure; (5) the data used for regression analysis is too scarce in quantity; (6) internal consistency of the data used for empirical calibration is not fully guaranteed; (7) the authors take energy parameters W’s (Margule’s parameters) independent of pressure and temperature, though the former are actually the functions of the latter items; and (8) the most important composition activity relationship is not consistent with the mineral compositions used for calibration. Furthermore, it is believed that empirical calibrations cannot avoid the following problems: (1) effect of retrograde metamorphism; (2) inconsistent data set; (3) effect of pressure on temperature estimation; and (4) effect of Fe 3+ calculation. Exact experimental work should be done in order to accurately calibrate the geothermometer in which the Margule’s parameters should be treated as the functions of pressure, temperature, and mineral compositions. This paper briefly introduces twenty one versions of garnet biotite Fe Mg exchange geothermometry, points out the sources of systematic errors inherent in geothermometry. The sources lie in that: (1) ideal Fe Mg mixing in garnet and biotite is assumed; (2) non ideal properties of both garnet and biotite are not considered completely; (3) minor elements-Mn, Ca in garnet, and Al VI, Ti, Mn in biotite are almost neglected; (4) effects of pressure on equilibrium, although little, are (5) the data used for regression analysis is too scarce in quantity; (6) internal consistency of the data used for empirical calibration is not fully guaranteed; (7) the authors take energy parameters W’s (Margule’s parameters) independent of pressure and temperature, though the former are actually the functions of the latter items; and (8) the most important composition activity activity is not consistent with the mineral compositions used for calibration. (1) effect of retrograde metamorphism; (2) inconsistent data set; (3) effect of pressure on temperature estimation; and (4 ) effect of Fe 3+ calculation. Exact experimental work should be done in order to accurately calibrate the geothermometer in which the Margule’s parameters should be treated as the functions of pressure, temperature, and mineral compositions.