Radiators represent the most spread heating body (installed since late 1800s) and in the last decades different radiators typologies have been proposed on the market, characterized by different materials, sizes, shapes, etc. Recent EU Directive on energy efficiency has set the obligation to install individual meters for space heating in building served by a central heating source. To this aim, when direct heat meters are not technically feasible, indirect systems like heat cost allocators are applied on each radiator in a dwelling and the knowledge of single radiators' thermal output is essential for an accurate and fair heat cost sharing. The EN 442:2014 describes a method for radiators' thermal output measurement whose expanded uncertainty is lower than 1% in reference laboratory conditions. However, radiators' thermal output is strongly dependent on installation and boundary conditions. Thus, to get radiators' thermal output at operating conditions "characteristic equations" are available but, unfortunately, they do not include any possible actual operating condition among which: installation position with respect to the wall and the floor, presence of grid/shelf/niche or an obstruction (e. g. caused by curtains), thermo-fluid-dynamic condition variations (inlet flow rate and temperature), and hydraulic connections. In this paper, the experimental results of thermal output measurement of different radiators typologies (cast iron, aluminum) at different installation conditions are presented, together with an analysis of the associate technical-economic effects on space heating cost sharing. Reductions of radiators' thermal output up to 15% due to hydraulic connections and between 10% and 20% due to flow-rate variations have been found. Furthermore, different installation conditions showed deviations between operating and standard radiators' thermal output between 5% and 15%.