Five groups of cases were selected for this study. All were tested at the Aural Rehabilitation Unit of the Deshon General Hospital between Nov., 1944, and May, 1945. These were divided into 5 groups: Group 1, 50 cases having flat audiograms characterized by equal loss in all frequencies; Group 2, 50 cases having gradual high tone loss characterized by a progressively greater impairment for higher frequencies at a slope of 5 to 10 decibels per octave; Group 3, 32 cases having marked high tone loss characterized by progressively greater impairment for higher frequencies at a slope of 15 to 20 decibels per octave; Group 4, 50 cases having notched audiogram characterized by flat or gradual high tone loss to 2048 c.p.s. and a sharp increased notch beyond 2048 c.p.s.; Group 5, 50 cases having atypical audiograms characterized by irregularities and curve shapes which exclude from any major category. These cases were selected because the audiogram presented the best example of the types found in the hundreds of cases going through at that time. 100 cases were also selected completely at random and the cases were also studied as a general group. The 7 groups described were compared on the basis of 3 criteria of auditory acuity: namely, speech reception threshold; ''better ear'' average for 512-2048 c.p.s.; and AMA percentage of loss measured. The data were treated by correlation methods and by analysis of variance. The following results and conclusions were obtained:[long dash]High positive correlation between speech reception and each of the 2 other criteria were found for all groups except those with marked high tone loss. Acuity for frequencies at the extremes of the audiometric test range is of minimal relationship to speech reception threshold. Patients with notches beyond 2048 c.p.s. cannot be differentiated on any important point from the group with flat losses. Group distributions for difference scores between speech reception threshold and pure tone average were not significantly differentiated except for cases of marked high tone loss. The latter showed a trend toward better score on speech reception than on pure tone average. The implication is that acuity between 512 and 1024 c.p.s. is more closely related to speech reception than is acuity between 1024 and 2048 c.p.s. Marked high tone loss cases are differentiated by sufficient factors so that this group may be accepted as constituting a special clinical category. The AMA % method gives less numerical parallelism with the speech measure than does the 512-2048 c.p.s. average. This fact, plus the extra labor involved, makes the pure tone average a more useful clinical tool except for special purposes. No statistical reason emerged from this study for favoring a percentage score over a simple ''better ear'' average.