Flatness of attenuation was measured for high-fidelity hearing protection devices (HPDs) using microphone-in-real-ear (MIRE) and behavioral real-ear-attenuation-at-threshold (REAT) protocols. In addition, participants completed the words-in-noise (WIN) test with and without HPDs, and provided subjective ratings of perceived sound quality via surveys administered after listening to music with and without HPDs. Attenuation was also measured using a manikin device, with both the manikin-based microphone and the MIRE probe tube used to make measurements inside the artificial ear coupler. Among human participants, statistically significant correlations revealed decreased WIN performance and poorer music quality ratings as achieved HPD attenuation increased. There was also a statistically significant correlation in which shallower slope (flatter attenuation) was correlated with higher perceived music quality ratings. Achieved attenuation decreased in the coupler when the probe tube was inserted into the manikin, suggesting HPD seals were compromised by the probe tube. Consistent with the manikin data, attenuation in human ears was greater when measured using REAT, suggesting similar compromise of the HPD seal during MIRE tests. To assure high fidelity HPDs are well fit, REAT testing is recommended to verify both overall attenuation and flatness of attenuation in individual listeners.