The paper presents the results of an extensive experimental research programme performed on safety valves in order to clarify the effects of back pressure and valve size on the flow capacity of a valve. As well known, back pressure strongly influences valve operating characteristics and can also reduce the discharge coefficient. This flow rate reduction can be related to the occurrence of a subsonic flow regime along the flow path and to insufficient disc lift. Last mentioned features can play a different role on different valve sizes because of the non exact geometrical scaling within the same valve size range. This happens because the requirements of typical application design standards, such as the API Standard 526, are such that the valve inlet, valve outlet and face to face dimension are not exactly scaled with respect to the orifice diameter. Moreover, face to face dimension can limit the body bowl volume leading to different device performances the same operational conditions. In order to clarify and evaluate the influence of the above mentioned parameters on the flow capacity of safety valves, many tests were carried on a single valve for different pressure ratios, disc lifts and for different valve outlet areas and body volumes representing different sizes derived from API Standard 526. Test results show significant differences on the flow capacity of safety valves under back pressure regime. This would suggest testing every valve size of the considered valve size range at different expansion ratios to confirm performance. Since this procedure leads to an excessive number of experimental tests, a sensitivity analysis on the influence of the most important geometrical parameters has been carried out. In order to minimize the number of experimental tests required for characterizing the flowing capacity of the whole valve size range, the paper proposes an experimental correlation for the prediction of the above mentioned non similarity effects.