In this work, a method for identifying counterfeit coins based on an electromagnetic acoustic transducer (EMAT) to detect the difference in the coin’s natural acoustic frequency response is presented. In the experimental system, the acoustic oscillation induced by a pulsed magnetic field is received by a microphone and recorded by an oscilloscope. The natural acoustic frequency of the coin is resolved by the fast Fourier transform (FFT) method on the computer. It is found that the natural frequencies of the possible counterfeit coins deviate significantly from the standard ranges of 16.9 to 17.4 kHz for the authentic 50 New Taiwan Dollar (NTD) coins. The observed natural frequencies of the coin are consistent with the values predicted by analytical estimation. We also built a prototype EMAT coin classification system to detect the natural acoustic frequency by direct frequency counting using a microcontroller. The prototype system demonstrates that a counterfeit coin can be identified by its natural frequency in less than 30 ms using the EMAT method. The proposed technique can be applied to the vending machine to improve the accuracy in discriminating between authentic and counterfeit coins.