This paper deals with the statistical modeling of key features of power line communication (PLC) channels that are necessary for designing data communication systems that operate over theses channels. The key features are average channel attenuation, root mean squared delay spread, coherence bandwidth and coherence time. All these features were estimated from in-home PLC channels measured in seven distinct Brazilian residences. Assuming that each feature is a random variable, four criteria (i.e., maximum likelihood estimate and three different information criteria) are used to select the statistical distribution that fits best to the data. The symmetry and asymmetry of the histogram associated with each feature is pointed out. The reported results focus on three frequency bands, namely: 1.7-30 MHz, 1.7-50 MHz and 1.7-100 MHz, which are in accordance with the standards used in Europe, North America and Brazil. The values obtained to Brazilian PLC channel features are different from those related to US and Europe channels reported in literature. Thus, the presented statistical models constitute an important tool to better design practical PLC systems that are suitable for Brazilian and in-home electric power grids.