Recently, a device called MammoSite, consisting of a balloon and a catheter, was developed to perform partial-breast irradiation using a high-dose-rate (HDR) brachytherapy unit with ease and reproducibility. However, the actual dose to the skin does not agree well with the calculated dose by the treatment planning system because of the difference between the calculation condition and the real treatment condition (i.e., homogeneous water and full scatter condition vs contrast solution and lack of full scatter condition). In this study, the authors experimentally estimated dose discrepancies due to contrast and lack of full scatter in breast HDR brachytherapy with MammoSite. Using metal-oxide-semiconductor field-effect transistor detectors and a breast phantom, the dose discrepancies between the calculation and the treatment conditions were measured according to contrast concentration (10% and 20% volume ratios), balloon size (35 and 60 cm3), and source to detector distance ranging from 25 to 50 mm. The source was an Ir-192 isotope from Nucletron HDR unit. The dose discrepancies from the calculation condition due to both contrast and lack of full scatter combined ranged from about -1.4 +/- 2.5% to -18.2 +/- 2.0% in the studied cases (error bound is in two sided confidence interval of 80% based on Student's t distribution). In all cases, the effect of lack of full scatter was dominant to that of contrast and significant dose discrepancies existed between the calculation and the real treatment conditions, indicating that the actual skin dose is less than that which is currently calculated.