A novel thermal measurement approach is presented for the determination of fan efficiencies. This approach is not resting on a simple direct measurement of the occurring temperature difference but on a system identification based on the transient time signal of the involved temperature measurement device. The values of the system parameters can be identified after recording the transient response of the system to a step input, i. e. the fan is suddenly turned on. By means of a representative case study considering a centrifugal fan, the system identification and its relation to fan efficiency are presented. The outcome of the new thermal approach is compared with results of standardized fan performance tests and with results obtained by the direct temperature measurement method without system identification. It is shown that the direct temperature measurement provides typically poor data whereas the novel thermal approach leads to results which are well comparable with these obtained by standard performance test at the design point.