The aim of this paper is to analyze the lower limit of detection (LLD), linearity of dose response, variation of radiation sensitivity between different tooth enamel samples, and time/temperature stability of EPR biodosimetry in tooth enamel. The theoretical LLD is shown to be 0.46 mGy, which is far lower than the measured value of about 30 mGy. The main issues to lowering LLD are the differentiation of the radiation-induced component against the total EPR spectrum and the complex nature of the dose dependence of the EPR signal. The following questions are also discussed in detail: need for exfoliated or extracted teeth from persons of interest, accounting for background radiation contribution; conversion of tooth enamel absorbed dose to effective dose; accounting for internal exposure specifically from bone-seeking radionuclides. Conclusions on future development of EPR retrospective biodosimetry are made.