In the present study a technique is developed to determine temperature profiles in hot glass melts, using intensity measurements performed at various wavelengths in the infrared spectrum. To that end an analytical model is developed which describes the internal energy transfer in a glass layer and the spectral intensity emerging from the glass layer. The spectral intensity so calculated is confronted with a measured spectral intensity to reconstruct the temperature profile in the glass layer. Because the temperature reconstruction from the measured spectral intensity is an ill posed inverse problem, Tikhonov regularization and the L-curve method are used to determine a meaningful temperature distribution in the glass layer. In order to gain insight into the accuracy of the temperature measurement method a sensitivity analysis is made using numerical simulations. The influence of random noise on the measurement signal and systematic errors in the properties is investigated. Furthermore, the temperature reconstruction of linear, logarithmic and parabolic temperature distributions is analysed.