How to Read a Visual Field: Hard-Core Perimetry Tips

Last updated: October 2026

A visual field printout contains everything you need to make a correct decision, if you read it in the right order. These are practical rules for reading Humphrey-style fields, spotting artifacts, and recognising glaucoma and neurological patterns.

The reading order

1. Patient details: name, eye, date of birth (age sets the normal values), date of test.
2. Test details: programme (24-2, 30-2, 10-2, 24-2C), strategy (SITA Standard, Fast, Faster), stimulus size (usually III), pupil size, and the corrective lens used.
3. Reliability indices and the gaze tracker.
4. Total deviation and pattern deviation probability plots.
5. Global indices: GHT, MD, PSD, VFI.
6. Greyscale, only as a quick visual summary.

Reliability: is this field worth reading?

• False positives above about 15% (SITA) make the field unreliable. Signs of a "trigger-happy" patient: white areas on the greyscale, impossibly high thresholds, and a positive MD that looks too good.
• Fixation losses above about 20% (blind-spot method) are a warning, but the blind spot method can be wrong if it was mapped incorrectly. The gaze tracker trace at the bottom is a better guide: a mostly flat line with few spikes is good.
• False negatives are often high in fields with real damage, so a high false-negative rate does not automatically mean poor reliability. It does mean you should be careful.
• The first field is often worse than later ones because of the learning effect. Never diagnose glaucoma from a single field; repeat it.

Total deviation vs pattern deviation

• Total deviation compares every point with age-matched normals.
• Pattern deviation removes any general depression and shows localised loss.
• Total deviation abnormal everywhere, pattern deviation normal = generalised depression: cataract, small pupil, uncorrected refractive error, or media opacity.
• Pattern deviation abnormal in a cluster = localised defect, the pattern that suggests glaucoma or a neurological cause.
• In very advanced glaucoma, pattern deviation can underestimate damage. Use total deviation and the raw thresholds.

Global indices

• GHT (Glaucoma Hemifield Test): compares matching zones above and below the horizontal meridian. "Outside normal limits" is one of the most useful signs of glaucomatous loss.
• MD (mean deviation): overall depression, strongly affected by cataract.
• PSD (pattern standard deviation): irregularity of the field, a good marker of localised loss. It paradoxically falls again in end-stage disease, when the whole field is depressed.
• VFI (visual field index): percentage of normal visual function, weighted towards the centre and less affected by cataract than MD.

Artifacts and mimics

• Lens rim artifact: a ring of depressed points at the edge of the test area, caused by the trial lens rim. Centre the lens close to the eye.
• Ptosis and heavy brows: superior depression. Tape the lid and repeat.
• Wrong correction: an uncorrected refractive error depresses the whole field. Use the correct age-appropriate near addition.
• Small pupil: pupils under about 3 mm reduce sensitivity. Dilate and repeat if needed.
• Fatigue: a "cloverleaf" pattern of better points around the four starting locations means the patient tired as the test went on.
• High myopia and tilted discs: enlarged blind spots and superotemporal defects that do not respect the vertical meridian, and can mimic glaucoma or chiasmal defects.

Glaucoma patterns

• Glaucomatous defects follow the nerve fibre layer and respect the horizontal meridian: nasal step, paracentral scotoma, arcuate (Bjerrum) defect, and later a residual central or temporal island.
• Match the field to the disc. Inferior rim thinning should produce a superior field defect, and vice versa.
• Central 10° is under-sampled: a 24-2 has only a few points in the central 10°. Patients with paracentral defects, NTG, or advanced disease need a 10-2 (or a 24-2C) to see what is happening near fixation.
• Minimum criteria for glaucomatous field loss (Hodapp-Parrish-Anderson): GHT outside normal limits, or PSD with p < 5%, or a cluster of at least 3 contiguous non-edge points at p < 5% on the pattern deviation plot in one hemifield, with at least one at p < 1%. Confirm on two consecutive reliable fields.
• Staging (Hodapp-Parrish-Anderson, by MD): early above −6 dB, moderate −6 to −12 dB, severe below −12 dB, with additional criteria for points near fixation.

Neurological patterns

• Defects that respect the vertical meridian mean chiasmal or retrochiasmal disease until proven otherwise.
• Bitemporal hemianopia: chiasmal compression (pituitary tumour). Superior defects first suggest compression from below.
• Junctional scotoma: a central scotoma in one eye plus a superotemporal defect in the other eye localises to the junction of the optic nerve and chiasm.
• Homonymous defects: the more posterior the lesion, the more congruous the defect. Macular sparing suggests an occipital cause. A superior "pie in the sky" defect suggests a temporal lobe lesion.
• Central and centrocaecal scotomas: optic neuritis, toxic or nutritional neuropathy, Leber hereditary optic neuropathy.
• Enlarged blind spot: papilloedema, peripapillary atrophy, or outer retinal disease such as MEWDS and AZOOR.

Detecting progression

• Establish a baseline with two reliable fields, ideally close together.
• Event analysis (Guided Progression Analysis): deterioration at the same 3 or more points on 2 consecutive tests is flagged as "possible progression", and on 3 consecutive tests as "likely progression".
• Trend analysis: the slope of VFI or MD over time (e.g., %/year or dB/year) shows how fast the patient is losing vision and helps predict lifetime risk.
• Expert guidance (including the World Glaucoma Association) recommends frequent testing in the first two years, roughly 3 fields a year, to identify fast progressors early.
• Cataract and cataract surgery change MD. Reset the GPA baseline after surgery.
• Always interpret progression together with OCT, IOP history and the clinical picture. See our perimetry algorithms page.