Differences in the glass transition temperatures (T-g) of epoxy resins were observed when their samples for differential scanning calorimetry (DSC) measurements were prepared by two different methods. The differences tin T-g) are believed to be due to the effects of interfacial residual stress, which is present only in the samples of lower T-g prepared with the formation of a "chemical" contact between epoxy resin and aluminum (pan) which have mismatched thermal expansion coefficients. Both types of the samples also showed, respectively, lower T-g when they were prepared by a fast-cooling process than when they were prepared by a slow-cooling process. This suggests that matrix residual stress, which is present in both types of the samples, as well as interfacial residual stress, has similar effects on the T-g of epoxy resins. This conclusion is substantiated by the results of the change of contact pattern and the freezing stress experiments.