This work deals with the specific studies of three main sources of uncertainty in electron spin resonance (ESR) dosimetry/dating of fossil tooth enamel: (1) the precision of the ESR measurements, (2) the long-term signal fading the selection of the fitting function. They show a different influence on the equivalent dose (DE) estimates. Repeated ESR measurements were performed on 17 different samples: results show a mean coefficient of variation of the ESR intensities of 1.20±0.23 %, inducing a mean relative variability of 3.05±2.29 % in the DE values. ESR signal fading over 5 y was also observed: its magnitude seems to be quite sample dependant but is nevertheless especially important for the most irradiated aliquots. This fading has an apparent random effect on the DE estimates. Finally, the authors provide new insights and recommendations about the fitting of ESR dose–response curves of fossil enamel with a double saturating exponential (DSE) function. The potential of a new variation of the DSE was also explored. Results of this study also show that the choice of the fitting function is of major importance, maybe more than the other sources previously mentioned, in order to get accurate final DE values.