Abstract: Purpose: In skin high-dose-rate (HDR) brachytherapy, sources are located outside, in contact with, or implanted at some depth below the skin surface. Most treatment planning systems use the TG-43 formalism, which is based on single-source dose superposition within an infinite water medium without accounting for the true geometry in which conditions for scattered radiation are altered by the presence of air. The purpose of this study is to evaluate the dosimetric limitations of the TG-43 formalism in HDR skin brachytherapy and the potential clinical impact. Methods: Dose rate distributions of typical configurations used in skin brachytherapy were obtained: a 5 cm x 5 cm superficial mould; a source inside a catheter located at the skin surface with and without backscatter bolus; and a typical interstitial implant consisting of an HDR source in a catheter located at a depth of 0.5 cm. Commercially available HDR Co-60 and Ir-192 sources and a hypothetical Yb-169 source were considered. The Geant4Monte Carlo radiation transport code was used to estimate dose rate distributions for the configurations considered. These results were then compared to those obtained with the TG-43 dose calculation formalism. In particular, the influence of adding bolus material over the implant was studied. Results: For a 5 cm x 5 cm Ir-192 superficial mould and 0.5 cm prescription depth, dose differences in comparison to the TG-43 method were about -3%. When the source was positioned at the skin surface, dose differences were smaller than -1% for Co-60 and Ir-192, yet -3% for Yb-169. For the interstitial implant, dose differences at the skin surface were -7% for Co-60, -0.6% for Ir-192, and -2.5% for Yb-169. Conclusions: This study indicates the following: (i) for the superficial mould, no bolus is needed; (ii) when the source is in contact with the skin surface, no bolus is needed for either Co-60 and Ir-192. For lower energy radionuclides like Yb-169, bolus may be needed; and (iii) for the interstitial case, at least a 0.1 cm bolus is advised for Co-60 to avoid underdosing superficial target layers. For Ir-192 and Yb-169, no bolus is needed. For those cases where no bolus is needed, its use might be detrimental as the lack of radiation scatter may be beneficial to the patient, although the 2% tolerance for dose calculation accuracy recommended in the AAPM TG-56 report is not fulfilled.