Smart condition monitoring and diagnosis of power system apparatus is an important element of the emerging Smart Grids, and in the context of high voltage, more attention has been focused on major and expensive equipment such as transformers. This study is concerned specifically with smart transformer monitoring. The problem of statistical indicators in the course of online transformer Frequency Response Analysis (FRA) evaluation is highlighted. Off-line and on-line applications of Frequency Response Analysis (FRA) are discussed in detail. A custom-made model transformer involving concentric LV and HV windings is fabricated and used as a test object. Its temperature and moisture content are changed in a controlled manner and their impacts on FRA responses are studied. To examine a real full-sized transformer, similar procedure is performed on a three-phase two windings core type transformer as another test object. Afterwards, statistical indicators are calculated using FRA data to clarify temperature and moisture influence on them. Deviation from the reference value for these indicators is discussed.