Following a nuclear incident, 131I monitoring is usually carried out by performing in vivo measurements of the retained activity in the thyroid. The measurement equipment needs to be calibrated in order to convert thyroid count rates into retained activity. This calibration relies on measurements or Monte Carlo computations using appropriate anthropomorphic model of the thyroid gland contaminated with a known activity of the radionuclide of interest. Several factors influence the calibration procedure, such as the geometry, the individual anatomical characteristics, and the presence of 131I in other organs and tissues of the body. In particular, given their age-dependent anatomical changes, appropriate values of detector efficiencies for children are necessary to reduce results uncertainties when children are involved. In this study, Monte Carlo simulations for several detectors and phantoms of different ages have been performed to investigate the variation of the detection efficiency with distance, age and thyroid volume. It was found that the counting efficiency varies linearly with the thyroid volume and inversely with the squared detector-phantom distance. Age-dependent correction factors to be applied to the adult calibration factor have been derived and can be used when calibration factors for children are not available.