Controversy presently exists concerning the relationship between loudness discomfort levels (LDL's) and acoustic reflex thresholds (ART's). In this study, LDL's and ART's were obtained for two groups of ten normal-hearing adult subjects and one group of ten adult subjects with bilateral sensorineural hearing losses. The procedure involved obtaining LDL's and ART's under earphones and under sound field conditions for four different acoustic stimuli: pure tones, warble tones, spondaic words, and speech spectrum noise. One group of normal-hearing subjects and the hard-of-hearing group of subjects were given "too loud, uncomfortably loud, or annoyingly loud" LDL instructions while a second group of normal hearing subjects were instructed to respond when a sound "first starts to become uncomfortable." Results indicated that LDL's, irregardless of instructional pattern, were reported at consistently higher sound pressure levels than the ART's, for all groups of subjects. The magnitude of the differences between the LDL's and ART's varied according to LDL instrucitons, type of test stimulus, hearing sensitivity of the subjects (normal hearing or hard-of-hearing) and transducer used for stimulus presentation. The results of this study demonstrate that ART's correlate too poorly with the LDL to permit ART's to be used as an objective measure of loudness discomfort.