How to Read Your Audiogram: What the Shape Says About Your Tinnitus

Most people are handed an audiogram, told a sentence about it, and leave with a piece of paper they cannot read. This is how to read it — and then the part that is specific to tinnitus, which is rarely explained at all and is genuinely interesting.

The chart

Two axes, and the vertical one is upside down relative to what you would expect.

Left to right: frequency, in hertz. Low pitches on the left, high pitches on the right. The standard test frequencies are 250, 500, 1000, 2000, 4000 and 8000 Hz, often with 3000 and 6000 added. Speech sounds mostly live between about 250 and 6000 Hz, which is why that is the range tested.

Top to bottom: decibels hearing level (dB HL), increasing downward. The further down the chart your mark sits, the louder a sound had to be before you heard it. Down is worse. This trips people up constantly, because every other chart they have seen puts good at the top.

The important thing about dB HL: it is not a measure of loudness in the ordinary sense. The zero point is a reference, not silence — 0 dB HL is the quietest level a young, healthy reference ear can detect at that frequency. Your threshold is expressed as the distance from that reference, which is why −5 dB HL is possible and simply means you did slightly better than average.

The symbols

• O — right ear, air conduction. Usually printed in red.
• X — left ear, air conduction. Usually printed in blue.
• Bracket-shaped marks (< > or [ ]) — bone conduction, tested with a small vibrator resting on the bone behind or in front of the ear.

Air conduction sends sound the whole way: ear canal, eardrum, the three middle ear bones, cochlea, auditory nerve. Bone conduction vibrates the skull directly, bypassing the outer and middle ear entirely.

Comparing the two is the single most informative thing on the chart:

• Air and bone together, both lowered — the problem is in the cochlea or beyond. Sensorineural. This is what noise damage and age-related loss look like.
• Bone normal, air lowered — an "air–bone gap" — sound is being blocked before it reaches a working inner ear. Conductive. Earwax, fluid in the middle ear, a perforated eardrum or otosclerosis produce this pattern, and most of them are treatable.
• Both lowered, with a gap as well — mixed.

If you take one thing from this section: an air–bone gap is the audiogram's way of saying something mechanical is in the way, and mechanical problems are the ones most likely to be fixable.

The numbers, and the fact that two authorities disagree

The pure tone average — usually the mean of your thresholds at 500, 1000 and 2000 Hz — is the single number most reports lead with, and the grade attached to it comes from a table.

The American Speech-Language-Hearing Association's table, in dB HL:

• Normal — −10 to 15
• Slight — 16 to 25
• Mild — 26 to 40
• Moderate — 41 to 55
• Moderately severe — 56 to 70
• Severe — 71 to 90
• Profound — 91 and above

The World Health Organization draws its lines elsewhere. It defines hearing loss as thresholds worse than 20 dB, and disabling hearing loss as greater than 35 dB in the better hearing ear.

Those two do not line up, and the mismatch is worth noticing rather than glossing. A threshold of 22 dB is normal by the WHO definition and a slight loss on the ASHA table. Neither is wrong; they are answering slightly different questions, and which word appears on your report depends partly on whose classification your clinic works from. If a number and a label seem to disagree, that is usually why.

One more thing the grade hides: it is an average across three low-to-mid frequencies. Someone with normal hearing at 500 and 1000 Hz and a substantial dip at 4000 Hz can have a pure tone average in the normal range and a clearly abnormal audiogram. The average is a summary, and the shape is the information.

The part that is about tinnitus

Here is what makes an audiogram worth reading properly if you have tinnitus, and it is not in most explanations of one.

A 2020 study in Otology & Neurotology reviewed 51,989 patients aged 18 to 80 who had an initial audiometric evaluation at Massachusetts Eye and Ear between 2000 and 2016. It asked which of them came in because of tinnitus, and what their audiograms looked like.

About 20% of everyone attending an initial hearing evaluation had tinnitus as their primary complaint. Its prevalence rose with age until roughly 50 to 54, then declined — not a straight line upward, which is itself worth knowing. Men were significantly more likely than women to present this way.

Then the finding about shape. Tinnitus as the primary complaint was statistically associated with specific audiogram configurations:

• Most prevalent with notched losses — a dip at one frequency band with better hearing either side of it, the classic pattern after noise exposure.
• Most prevalent with steeply sloping high-frequency losses.
• Relatively uncommon in adults with flat losses — thresholds lowered roughly evenly across all frequencies.

The kind of tinnitus reported tracked the shape too. Patients with frequency-restricted threshold shifts often described tonal percepts — a defined pitch. Patients with asymmetric configurations tended to describe broadband percepts — hissing, rushing, static rather than a note.

The authors read this as support for a specific mechanism: that tinnitus arises from sharp discontinuities in the pattern of surviving nerve supply and cochlear amplification, which then induce topographically restricted changes in the central auditory pathway. Not the amount of damage. The edge of it.

That is why a notch matters more than its depth suggests. A cliff in the audiogram is a cliff in the map the brain has of frequency, and the mechanism described throughout this site — reduced input, compensating central activity — has its sharpest expression right at that boundary. It is also, incidentally, the reasoning behind notched sound therapy, which targets the frequency band rather than masking everything — an elegant idea whose trials have struggled to show that the notch is doing the work.

A caution about reading your own chart against this. This is a retrospective study of who walked into one clinic, so it describes association, not a rule about individuals. Plenty of people have flat losses and severe tinnitus. What it gives you is a reason the shape is worth looking at, not a prediction.

If you have the numbers and would rather not eyeball the shape, the audiogram reader plots them, works out the average and the grade on both scales, and names the configuration — with the criteria it used shown beside the answer, because there is no standard definition of a notch to appeal to. It runs entirely in your browser.

Asymmetry, and the hyperacusis finding

A 2023 study in Hearing Research took 626 adults with documented hyperacusis and compared them against an age- and sex-matched group from the same clinic who did not report it.

The result is counterintuitive. Patients with hyperacusis had better high-frequency thresholds from 2000 to 8000 Hz than the comparison group — but significantly larger differences between their two ears across 250 to 8000 Hz. The probability of reporting hyperacusis was highest for normal, asymmetric and notched configurations.

The authors' interpretation is that a central compensatory mechanism dominated by input from the intact or less damaged ear may overshoot, raising neural activity past its baseline and producing sensitivity rather than loss.

The practical version: if you have hyperacusis, a normal-looking audiogram is not evidence against you, and the number worth looking at is the difference between your ears rather than either ear's absolute thresholds. Many in that group reported one-sided symptoms, a history of noise exposure, and tinnitus alongside.

What the audiogram does not measure

An audiogram answers one question well: what is the quietest pure tone you can detect, in a quiet room, at these particular frequencies. Several things fall outside that.

• Anything above 8000 Hz. Standard audiometry stops there. Damage above the tested range is invisible to it, and noise damage often begins high.
• Hearing in noise. Detecting a tone in silence and following a conversation in a restaurant are different tasks, and people fail the second with a normal result on the first. The digits-in-noise test probes that separately.
• The connections between hair cells and the auditory nerve. These can be lost while thresholds stay normal, which is the subject of hidden hearing loss.
• Your tinnitus itself. Nothing on the chart shows it. How tinnitus is measured covers the instruments that do, and why they measure distress rather than sound.

A normal audiogram with tinnitus is common and is not a contradiction. It means the test found nothing in the range it tests.

What to do with this

• Ask for a copy. You are entitled to it, it is one page, and it is the baseline every future test gets compared against. Getting the first one is the only way to know later whether anything has changed.
• Look at the shape before the grade. A notch or a steep slope carries information the pure tone average averages away — and on the 51,989-patient data, shape is what tracks with tinnitus.
• Check the gap. Bone conduction better than air conduction points to a mechanical cause, and mechanical causes are the treatable ones.
• Compare your two ears. A meaningful asymmetry is worth asking about, both because it associates with hyperacusis and because one-sided findings get investigated differently.
• If it came back normal, that is a real result and not a dismissal. What a first audiology appointment involves covers what usually follows, and being told nothing is wrong covers what to do when the conversation stops there.