An experimental investigation of three couplers, suitable for testing circumaural earphones, was made to determine the limitations on their use. Two of these couplers were modified forms of standard U.S. couplers, the ASA‐type 1 and the NBS‐9A, while the third was a simple flat plate with a hole in which the test microphone is flush mounted. Results presented are based on tests made using the Telephonics TDH‐39 earphone with NAF‐48490‐1 cushion and the Philips model HA‐10 circumaural earphones. Below 2000 Hz, all three couplers are reliable and useful for measuring the responses of circumaural earphones. Above 2000 Hz, small variations in the electroacoustic system, defined by the earphone, cushion, coupler, and measurement microphone, cause large variations in the measured response. Variations in earphone responses are shown to be caused by variations in (1) earphones of a particular type, (2) cushions of a particular type, (3) applied force on the earphone, and (4) placement of the earphone on the coupler. These variations are in addition to the expected variations in the measured responses of different types of earphones when used with different types of cushions and when measured on different types of couplers.