At the Bhabha Atomic Research Centre, a thin (76 mm diameter x 2 mm thickness) NaI (Tl) detector is used for the assessment of ¹²⁵I in the thyroid of the radiation workers engaged in the preparation of radio-immunoassay kits. The detector was calibrated using a REMCAL (radiation equivalent manikin calibration) phantom with a known amount of the ¹²⁵I activity filled in its thyroidal cavity. Since ¹²⁵I emits low-energy photons ranging from 27 to 35.4 keV, its detection efficiency depends on several parameters such as neck-to-detector distance, detector size, unknown tissue thickness overlying (OTT) the thyroid and the shape and size of the thyroid. To account for uncertainties introduced by these factors in the estimation of ¹²⁵I, a computer program based on the Monte Carlo photon transport techniques was developed. The program simulates the detector response and the corresponding detection efficiencies using two thyroid models: (1) revised MIRD head phantom and (2) Ulanvosky model. The program has been validated with experimental measurements carried out using a REMCAL phantom. The computed values of uncertainties due to placement errors (+0.5 cm) for different detector sizes, differences in the OTT of the thyroid (0.6–2.0 cm) and different thyroid shapes are presented in this paper. The computed values of the calibration factors, determined for the revised MIRD phantom, varied from 5.23 to 1.06 x 10^-2 counts per photon for detector distance of 3–12 cm and from 7.53 to 3.66 x 10^-2 counts per photon for OTT varying from 0.6 to 2.0 cm keeping the detector at a distance of 3 cm. This study shows that the variations in OTT constitute a major source of uncertainty. The computed uncertainties due to various parameters should be taken into account while estimating the thyroidal burden of ¹²⁵I in the radiation workers. The feasibility of using coincidence method for absolute determination of the ¹²⁵I activity in the thyroid is also discussed in this paper.