Breast tissue is heterogeneous and is mainly composed of glandular (G) and adipose (A) tissues. The proportion of G versus A varies considerably among the population. The absorbed dose distributions in accelerated partial breast irradiation therapy with low energy photon brachytherapy sources are very sensitive to tissue heterogeneities. Current clinical algorithms use the recommendations of the AAPM TG43 report which approximates the human tissues by unit density water. The aim of this study is to investigate various breast tissue modeling schemes for low energy brachytherapy. A special case of breast permanent seed implant is considered here. Six modeling schemes are considered. Uniform and non-uniform water breast (UWB and NUWB) consider the density but neglect the effect of the composition of tissues. The uniform and the non-uniformG/Abreast (UGAB and NUGAB) as well the age-dependent breast (ADB) models consider the effect of the composition. The segmented breast tissue (SBT) method uses a density threshold to distinguish between G and A tissues. The PTV D(90) metric is used for the analysis and is based on the dose to water (D(90(w),(m))). D(90(m,m)) is also reported for comparison to D(90(w,m)). The two-month post-implant D(90(w,m)) averaged over 38 patients is smaller in NUWB than in UWB by about 4.6% on average (ranging from 5% to 13%). Large average differences of G/A breast models with TG43 (17% and 26% in UGAB and NUGAB, respectively) show that the effect of the chemical composition dominates the effect of the density on dose distributions. D(90(w,m)) is 12% larger in SBT than in TG43 when averaged. These differences can be as low as 4% or as high as 20% when the individual patients are considered. The high sensitivity of dosimetry on the modeling scheme argues in favor of an agreement on a standard tissue modeling approach to be used in low energy breast brachytherapy. SBT appears to generate the most geometrically reliable breast tissue models in this report.