In Brazil, the Electricity Regulatory Agency defines that technical losses on distribution systems must be calculated via power flow simulation, with primary and secondary systems modeled in detail. However, as this approach requires a massive amount of data not always available in practice, the agency established assumptions to enable its implementation by all Brazilian utilities. These assumptions are currently under scrutiny by the utilities. In this context, this paper investigates the assumptions regarding load connection type, load power factor, line impedance, grounding resistance and load curve resolution, to identify whether they are indeed necessary, technically sound and what is the quantitative impact on technical losses. The study was conducted on a large database with 1,248 distribution feeders modeled in OpenDSS (over 4.5 million customers) to obtain truly realistic results. Moreover, analytical derivations are presented to assess the technical correctness of all assumptions. It is found that, for the studied feeders, the assumptions seem to be reasonable, but more studies are necessary to define the load power factor and how to model reactive power compensation equipment.