Comparative analysis of several methods for purity determination using DSC is presented. This is based on a mathematical model including the construction of theoretical melting curves for two-component systems and the calculation of recorded melting curves with the help of a set of equations describing the formation ofa DSC output signal. It is shown that the true accuracy of purity determinations in the range of impurity concentrations ~ = 0.005-0.02 does not exceed 30-50%.Purity determination is a very valuable and important application of DSC that has been subject of several reviews [2], [6][7][8][9][10], but the problem of establishing the true accuracy of DSC purity determinations has not yet been solved. This was convincingly demonstrated in [7] and [8], where it was shown that the relevant publications include obviously overstated estimates of accuracy as well as recommendations to apply DSC only for determination of the order of the impurity concentration.The essence of the problem may be formulated as follows: for an isobaric melting process of a two-component eutectic system in thermodynamic equilibrium, function 7" = ~ , where F is the fraction of the liquid phase in the system at temperature T, should be a straight line with slope proportional to the impurity concentration and cutting off the value of the melting temperature of the pure main z~x component on the ordinate. However, functions T = J( F) constructed on the X--/ basis of the experimental melting curves obtained with DSC turn out to be essentially nonlinear. The absence of thermodynamic equilibrium in a dynamic experiment, the inability of DSC to register the initial stages of melting, the formation of a solid solution, and the influence of thermal gradients, have all been indicated as possible causes of this nonlinearity. In order to eliminate the