Comfortable loudness judgments for discrete frequency signals.

The test-retest reliability of Most Comfortable Loudness Level (MCL) was examined in three experiments using young normal-hearing adults unsophisticated in audiometric testing. Test stimuli included pure tones, narrow-band noise, and damped wave trains, with presentation by ascending, descending, and Bekesy procedures. Intersubject reliability was found to be very poor and intrasubject reliability to be only fair. Although instructions to Ss and subjects interpretation of instructions probably affect MCL reliability, the task appears to be inherently variable.