Results comparing hearing aid performance utilizing probe microphones with performance measured in a commercial hearing aid test box using an HA-2 coupler are presented. Sixteen hearing-impaired subjects (23 ears) were divided into two groups. Group I utilized unvented earmolds with short canals and large bores; group II utilized unvented earmolds with long canals and small bores. Results revealed that large individual differences across frequencies as well as within and between groups preclude predicting how a hearing aid will perform on a real ear from test box measurements.