Background and purpose Refractometers have gained a foothold in childhood screening for ophthalmic disorders. Given the results of an ophthalmic follow-up of an extremely preterm Danish cohort, the results of the Retinomax autorefractor were further evaluated. Materials and methods A nationwide cohort of infants born before gestational age 28 weeks (n = 178) and 56 term controls were examined at the age of 4 years. Refraction was given as the cycloplegic Retinomax value. For this study, we analysed the equipment's confidence value on the printout and equipment-induced myopization (as the difference between refraction measured before and after topical cyclopentholate 1%), both items hypothetical with a view to having identified factual ophthalmic deviations. Results Thirty-two of 42 eyes with visual acuity ≤ 0.4 had high Retinomax confidence values (8-9); the Retinomax values were also high in 10 of 12 children with strabismus and lack of stereopsis. Low values (1-6) were recorded in 11 single eyes, 5 of which were normal (false positives). Three children already known to have low vision were unable to cooperate. The overall mean value for equipment-induced myopization was 1.9 D (range, 0-6.87 D). Myopization showed no correlation with visual acuity and corneal curvature, and a weak positive correlation with refractive value disappeared when the myopic outliers were excluded. Conclusions The hand-held Retinomax seemed to be reliable for assessing refraction in 4-year-old children, provided a cycloplegic