Stereoacuity Tests Compared: Randot, TNO, Titmus, Frisby, Digital and VR

Updated 2026-10-22

Written by the Optics Trainer team from published research, with every source linked below.

Stereoacuity is quick to measure, but the number depends heavily on which test you use. Clinical stereotests separate the two eyes' images in different ways, carry different monocular cues, and give different thresholds in the same patient. This guide compares the common book tests, explains how much change between visits is real, summarises published values for children, and covers what digital and VR stereotests show so far.

How the common tests separate the eyes

Stereotests present a slightly different image to each eye in one of several ways, as Read's 2015 review describes:

Polarised vectographs: Titmus (contour circles) and Randot (contour and random-dot targets), viewed through polarised glasses.
Red/green anaglyph: the TNO random-dot test.
Lenticular cards: the Lang test, which needs no glasses.
Real depth: the Frisby plates, where the target is printed at a different physical depth from its surround.

Monocular cues

Contour tests such as the Titmus circles can in principle be passed with monocular cues, because the contour shifts sideways, and a monocular re-test is one way to check. Random-dot and real-depth tests reduce contour cues, but the Frisby and Lang tests can still offer motion-parallax cues if the patient moves their head.

Why tests give different results

Different stereotests give different thresholds in the same patients and are not interchangeable. Patients with strabismus often reach finer thresholds on real-depth Frisby plates than on Randot, and on contour tests than on random-dot tests.

The TNO test has agreed poorly with other tests, and in one study of 149 children aged 4 to 16 its thresholds were about twice as high. Proposed reasons include its protocol and disparities and the red/green anaglyph glasses (Vancleef 2017; a 32-person lab study by Zhang 2021); neither proves a single cause.

How much change is real

In 36 patients with stable strabismus, Adams and colleagues (2009) found a change of about two octaves, two doublings of threshold, was needed before a change on most of the tests they studied exceeded test-retest variability. A one-level change on a stereo plate may be noise. Ceiling effects also limit how well common tests separate people with good stereo vision.

Published values in children

Values are specific to the test and the age group. In 4,355 children with normal vision, Randot Preschool mean stereoacuity improved from 100 arcsec at age 3 to 40 arcsec by age 7 (Birch 2008). A later study of the same test found most children testable from age 3 and that it reliably separates any stereo from none, but agreement on the exact threshold in children aged 2 to 5 was poor and most children over 5 reached its best score.

Stereopsis in amblyopia and strabismus

Stereopsis is more affected in strabismic than in anisometropic amblyopia. In a pooled PEDIG cohort of 633 children with anisometropic amblyopia, 37% had no measurable stereoacuity at baseline, and many remained below age-typical levels even after their visual acuity improved. A stereotest alone does not diagnose amblyopia or strabismus.

Digital and VR stereotests

Newer tests use screens and headsets:

ASTEROID is a glasses-free random-dot test on an autostereoscopic tablet with an adaptive staircase. In young adults, version 1.0 gave thresholds similar to Randot Circles.
In 112 children, ASTEROID and Titmus agreed only moderately, and ASTEROID usually gave the poorer result when they disagreed. The authors note it is unknown which test is more accurate.
A VR headset contour stereotest (Vivid Vision) correlated weakly with ASTEROID in 64 healthy young adults and had poor-to-moderate repeatability.

Choosing and recording a stereotest

Because tests are not interchangeable, the most useful habit is consistency: use the same test, distance and lighting at each visit, record which test was used, and judge change against the roughly two-octave change Adams found for most tests in strabismus rather than a single level. In a 2019 survey of eye care professionals, Frisby was the most used test in the British Isles, and Titmus and Randot were more common in the US and Canada.

How Optics Trainer fits

Optics Trainer VR's activity library includes stereo acuity activities, and Optics Tracker includes Worth 4 Dot testing for fusion and suppression. None of the studies above evaluated Optics Trainer, and these activities are not presented as validated stereotests.

Optics Tracker is clinical decision-support software. Its results support, and do not replace, standard clinical testing, professional diagnosis and clinical judgement.

Frequently asked questions

Which stereoacuity test is most accurate?

None is established as most accurate, and tests are not interchangeable. They separate the eyes differently, carry different monocular cues, and give different thresholds in the same patient. Use the same test each time.

Why does the TNO test give worse stereo scores?

TNO thresholds have been about twice those of other tests in one study of children. Proposed reasons include its protocol and disparities and its red/green anaglyph glasses, but no single cause has been proven.

How much stereoacuity change is meaningful?

About two octaves, or two doublings of the threshold, on most of the tests Adams and colleagues studied in patients with stable strabismus. Smaller changes may be test-retest noise.

What is normal stereoacuity for a child?

It depends on the test and age. In 4,355 children with normal vision, Randot Preschool mean values were 100 arcsec at age 3, improving to 40 arcsec by age 7. Use values for the specific test you are using.