A new protocol for testing on-grid inverters accounting for the operating conditions specific to CPV arrays is introduced. First, a CPV-specific weighted average for providing the static conversion efficiency according to typical frequency distributions of CPV yield throughout the year. Then, a set of representative scenarios for analyzing the dynamic MPPT efficiency of the inverter: slow ramp (clear sky), fast cloud passage, self-shading between arrays (typical of sunrise and sunset) and a wavering current response typical of imperfect tracking. The procedures have been applied to two different string-size inverters at the CEA inverter test bench, which have revealed strong differences in their dynamic MPPT efficiency. The CPV-specific weighted average proposed was found to provide lower performance figures than the classical European and CEC efficiencies for the two studied inverters. This protocol provides important information for CPV plant dimensioning and equipment selection, as the differences in performance revealed can result in significant deviation on the final energy yield.