Aim To establish the range of normal stereoacuity thresholds and evaluate the diagnostic reliability of stereoacuity tests in preschool-aged children.

Methods 1606 children, aged 24–72 months, had detailed eye examinations and stereoacuity testing. Lang-Stereotest II (LangII) was attempted on all children, Stereo Smile Stereoacuity II Test (SSST) was conducted on children aged <30 months and on older children who could not complete the Randot Preschool Stereoacuity Test (RPST). The RPST was conducted on children aged ≥30 months and on some younger children who passed both the LangII and SSST.

Results Modes for the age groups 24–47 months and 48–72 months were: 200 arcsec for both age groups with the LangII test; 120 arcsec and 60 arcsec, respectively, with the SSST; 100 arcsec and 60 arcsec, respectively, with the RPST. Age-adjusted areas under the curve for detecting amblyopia, strabismus and anisometropia were: for the LangII test, 0.72, 0.68 and 0.60, respectively; for the SSST, 0.73, 0.80 and 0.57, respectively; for the RPST, 0.92, 0.82 and 0.73, respectively.

Conclusions Normative data for the LangII, RPST and SSST stereoacuity tests were determined for children aged 24–72 months. Sensitivity and specificity at individual disparity levels for detecting anisometropia, amblyopia and strabismus were also determined for RPST and SSST. Using area under age-adjusted receiver operating curves, the RPST was found to be the most reliable in detecting ocular conditions compared with the LangII and SSST tests.