In this paper the use of the photothermal radiometry technique in the frequency domain (PRTF) and the use of multivariate methods in the study of two types of skin phantoms: (a) one in which skin pigmentation was simulated dyeing the gel phantom and (b) the other consists of exposure of animal skin samples to different degrees of thermal damage. In experiment (a), gel phantoms were prepared with different concentrations of methylene blue (MB). The mean values of the radiometry signal (RS) show significant differences in only those cases in which changes in the concentration of MB were higher than 0.38 mM. This result was confirmed with a t test for independent samples of the data ( p < 0.05). The mean values of the amplitude and phase signal do not permit discrimination between phantoms with changes in pigmentation equal to or lower than this value. However, principal component analysis (PCA) demonstrated that it is possible to discriminate between phantoms with changes in molar concentration equal to 0.38 mM (for the phase signal). In the case of experiment (b), the following four groups of pork skin samples were analyzed: one consists of samples of fresh skin, while the other three consist of samples exposed to thermal damage at 45 • C (the exposure time was 4 s) and 80 • C (exposure times were 4 s and 8 s, respectively). The mean values of the RS for each group of samples did not show a clear visual discrimination. However, the t test for independent samples applied to the data demonstrated significant differences only between fresh skin and skin exposure to thermal damage at 80 • C (with exposure times of 4 s and 8 s). PCA was used to discriminate between the four different skin samples.