In case of radiological accident, retrospective dosimetry is needed to reconstruct the absorbed dose of overexposed individuals not wearing personal dosimeters at the onset of the incident. In such a situation, emergency mass triage will be required. In this context, it has been shown that Electron Paramagnetic Resonance (EPR) spectroscopy would be a rapid and sensitive method, on the field deployable system, allowing dose evaluation of a great number of people in a short time period. This methodology uses tooth enamel as a natural dosimeter. Ionising radiations create stable free radicals in the enamel, in a dose dependent manner, which can be detected by EPR directly in the mouth with an appropriate resonator. Teeth are often subject to restorations, currently made of synthetic dimethacrylate-based photopolymerizable composites. It is known that some dental composites give an EPR signal which is likely to interfere with the dosimetric signal from the enamel. So far, no information was available about the occurrence of this signal in the various composites available on the market, the magnitude of the signal compared to the dosimetric signal, nor its evolution with time. In this study, we conducted a systematic characterization of the signal (intensity, kinetics, interference with dosimetric signal) on 19 most widely used composites for tooth restoration, and on 14 experimental resins made with the most characteristic monomers found in commercial composites. Although a strong EPR signal was observed in every material, a rapid decay of the signal was noted. Six months after the polymerization, the signal was negligible in most composites compared to a 3 Gy dosimetric signal in a tooth. In some cases, a stable atypical signal was observed, which was still interfering with the dosimetric signal.